Ilakiyaselvan N., Nayeemulla Khan A., Shahina A.. Deep learning approach to detect seizure using reconstructed phase space images[J]. The Journal of Biomedical Research, 2020, 34(3): 240-250. DOI: 10.7555/JBR.34.20190043
Citation:
Ilakiyaselvan N., Nayeemulla Khan A., Shahina A.. Deep learning approach to detect seizure using reconstructed phase space images[J]. The Journal of Biomedical Research, 2020, 34(3): 240-250. DOI: 10.7555/JBR.34.20190043
Ilakiyaselvan N., Nayeemulla Khan A., Shahina A.. Deep learning approach to detect seizure using reconstructed phase space images[J]. The Journal of Biomedical Research, 2020, 34(3): 240-250. DOI: 10.7555/JBR.34.20190043
Citation:
Ilakiyaselvan N., Nayeemulla Khan A., Shahina A.. Deep learning approach to detect seizure using reconstructed phase space images[J]. The Journal of Biomedical Research, 2020, 34(3): 240-250. DOI: 10.7555/JBR.34.20190043
N. Ilakiyaselvan, School of Computer Science and Engineering, Vellore Institute of Technology, Vandalur-Kelambakkam Road, Chennai, Tamilnadu 600127, India. Tel/Fax:+91-44-39931286, E-mail: ilakiyaselvan.n@vit.ac.in
Epilepsy is a chronic neurological disorder that affects the function of the brain in people of all ages. It manifests in the electroencephalogram (EEG) signal which records the electrical activity of the brain. Various image processing, signal processing, and machine-learning based techniques are employed to analyze epilepsy, using spatial and temporal features. The nervous system that generates the EEG signal is considered nonlinear and the EEG signals exhibit chaotic behavior. In order to capture these nonlinear dynamics, we use reconstructed phase space (RPS) representation of the signal. Earlier studies have primarily addressed seizure detection as a binary classification (normal vs. ictal) problem and rarely as a ternary class (normal vs. interictal vs. ictal) problem. We employ transfer learning on a pre-trained deep neural network model and retrain it using RPS images of the EEG signal. The classification accuracy of the model for the binary classes is (98.5±1.5)% and (95±2)% for the ternary classes. The performance of the convolution neural network (CNN) model is better than the other existing statistical approach for all performance indicators such as accuracy, sensitivity, and specificity. The result of the proposed approach shows the prospect of employing RPS images with CNN for predicting epileptic seizures.
Postmenopausal osteoporosis (PMOP) is a bone metabolic disease that results in reduced bone mass and the destruction of bone tissue microstructure[1]. It primarily affects middle-aged and elderly women, representing the most prevalent form of osteoporosis in this demographic. Statistically, approximately 50% of women sustain at least one fracture after menopause[2]. Because of the aging global population, there has been a significant rise in the number of patients suffering from postmenopausal osteoporosis, resulting in a substantial demand for medical resources. PMOP is caused by the decline of ovarian function after menopause and the sharp drop in estrogen levels in the body, leading to an imbalance in bone remodeling. Bone remodeling is a strictly regulated process that involves bone resorption mediated by osteoclasts and bone formation mediated by osteoblasts, with a dynamic balance between bone resorption and bone formation[3–4]. Consequently, it is of paramount importance to elucidate the precise mechanisms underlying bone remodeling after PMOP. The bilateral ovariectomy (OVX) mouse model, a classical model for simulating postmenopausal osteoporosis, is extensively used to analyze the mechanism of estrogen-deficient osteoporotic diseases[5]. First, OVX was found to affect the stabilization of β-catenin through the down-regulation of the classical Wnt/β-catenin pathway, which in turn downregulated the expression of both runt-related transcription factor 2 (RUNX2) and osterix (OSX) that were involved in osteoblast differentiation, maturation, and activity[6–7]. Likewise, OVX also activated the NF-κB pathway in osteocytes and induced osteocyte apoptosis[8]. Second, OVX enhanced osteoclast differentiation and accelerated bone resorption in mice[9–10]. The increased activity of osteoclasts induced by OVX occurred through the receptor activator of nuclear factor-kappaB (NF-κB) ligand (RANKL)-mediated NF-κB and nuclear factor of activated T-cells 1 (NFATc1) pathways[11]. The RANKL was reported to bind to the receptor activator of NF-κB (RANK) and then activate the key osteoclast regulators, such as NFATc1, c-Fos, and CTSK, to promote the osteoclast differentiation[12–14]. Third, the differentiated osteoblasts also produced both positive and negative regulators of osteoclastogenesis, including RANKL and its natural decoy receptor osteoprotegerin (OPG)[15]. Estrogen deficiency upregulated the expression of RANKL but downregulated the expression of OPG, resulting in an increase in the RANKL/OPG ratio[16]. Fourth, the withdrawal of estrogen was associated with a spontaneous increase in pro-inflammatory cytokines, such as TNF-α, IL-1β, and IFN-γ, that stimulated bone resorption[17]. Overall, OVX hinders osteogenesis and facilitates the development of osteoclastogenesis, leading to bone loss.
Differentiated embryo-chondrocyte expressed gene 1 (DEC1) is a structurally unique basic helix-loop-helix protein that is a major regulator of a variety of physiological and pathological processes, including circadian rhythms, hypoxia, cellular proliferation, apoptosis, immunity, and inflammation[18–22]. Initially identified in human chondrocytes, DEC1 is closely linked to bone growth and remodeling[23]. We previously found that DEC1 enhanced osteoblast osteogenesis and that icariin, which upregulated DEC1 expression in SaoS2 cells, might protect against glucocorticoid-induced osteoporosis[24]. These studies imply that DEC1 is a crucial factor in the formation of osteoblasts. Furthermore, our previous study also found a decrease in DEC1 in the bone tissue of OVX mice[25]. However, the role of DEC1 in PMOP remains to be determined.
In the current study, we used two genotypes (Dec1+/+ and Dec1−/−) of mice to establish an OVX model and determine the effects and mechanisms of DEC1 in PMOP.
Materials and methods
Animals
The heterozygous Dec1 (Dec1+/−) C57BL/6 mice were purchased from the RIKEN BioResource Research Center in Japan and bred at the Animal Core Facility of Nanjing Medical University. Dec1 heterozygous (Dec1+/−) C57BL/6 mice were mated to obtain offspring, from which Dec1 gene knockout (Dec1−/−) mice and wild-type (Dec1+/+) mice were selected from the littermates. Double checks (after birth and before the experiment) were applied to ensure the correct mouse genotype. The mouse genotype was confirmed as shown in Supplementary Fig. 1A (available online). Mice were housed in a specific pathogen-free barrier environment with good ventilation, a 12 h∶12 h light-dark cycle, temperature maintained at 22–26 ℃, and humidity kept constant at 50% to 60%, with free access to water and food. Efforts were made to minimize animal suffering and to reduce the number of animals used for the experiments. All the animal experiments were strictly in compliance with the experimental animal guidelines of the Laboratory Animal Research Institute and were approved by the Animal Ethical and Welfare Committee of Nanjing Medical University (IACUC-2203015).
Establishment of ovariectomy mouse models
OVX mouse models were established according to a previously reported study with minor modifications[26]. Both 20 Dec1+/+ mice and 20 Dec1−/− mice, aged 6–8 weeks, were randomly divided into the sham and OVX groups, with 10 mice in each group. After intraperitoneal injection of 4% chloral hydrate (0.01 mL/10 g) for anesthesia, the mice in the OVX group were ovariectomized bilaterally, and the corresponding control mice were sham-operated (sham).
BMSCs and BMMs were isolated and cultured from the OVX mice
One month after the operation, the mice of Dec1+/+ sham, Dec1+/+ OVX, Dec1−/− sham, and Dec1−/− OVX were euthanized, and the tibiae and femurs were aseptically dissected to collect bone marrow cells. The adherent cells were cultured until reaching a confluence of 80%. BMSCs from passages five to nine were harvested for subsequent experiments. BMMs from the mice were cultured in alpha-MEM medium (Cat. #12571071, Gibco, Grand Island, NY, USA) supplemented with 5 ng/mL macrophage colony-stimulating factor (M-CSF; Cat. #315-02, PeproTech, Rocky Hill, NJ, USA), 1% penicillin/streptomycin, and 10% fetal bovine serum (FBS; Cat. #086-150, Wisent, Saint-Jean-Baptiste, QC, Canada). Both BMSCs and BMMs from the two groups of mice (Dec1+/+ sham and Dec1+/+ OVX) or four groups of mice (Dec1+/+ sham, Dec1+/+ OVX, Dec1−/− sham, and Dec1−/− OVX) were used in subsequent experiments.
Micro-CT analysis
The femurs of mice were fixed overnight in 4% paraformaldehyde and subjected to the micro-CT analysis using a SkyScan scanner (SkyScan1172, Bruker, Kontich, Belgium) with a scanning precision of 18 μm. The obtained scan data were reconstructed in three dimensions using the CTvox software. Bone trabecular and cortical bone parameters, including bone mineral density (BMD, mg/cc), bone volume fraction (BV/TV), trabecular number (Tb.N, per mm), trabecular thickness (Tb.Th, μm), trabecular separation (Tb.Sp, mm), total area (Tt.Ar, mm2), cortical bone area (Ct.Ar, mm2), and cortical bone thickness (Ct.Th, mm), were analyzed.
Morphometric analysis of bone tissues
The femurs of five-month-old mice were fixed in 4% paraformaldehyde for 24 h, followed by decalcification in 10% EDTA for two weeks. Subsequently, the bones were dehydrated in ethanol and xylene, embedded in paraffin, and sectioned into 5-μm thick slices using a paraffin microtome (RM2245, Leica, Wetzlar, Germany). The bone sections were stained with hematoxylin and eosin (HE) staining (Cat. #D006-1-4, Nanjing Jiancheng Bioengineering Institute, Nanjing, Jiangsu, China).
Immunohistochemistry (IHC) staining
Sections were dewaxed in xylene and then rehydrated through an ethanol concentration gradient. Tissue samples were then deactivated with 3% H2O2. Antigen retrieval was performed using trypsin (Cat. #T8150, Solarbio, Beijing, China), followed by blocking the sections with 5% goat serum (Cat. #AR0009, BOSTER, Pleasanton, CA, USA) for one hour. Subsequently, sections were incubated overnight at 4 ℃ with specific antibodies against β-catenin (1∶500, Cat. #PK02151, Abmart, Shanghai, China), RUNX2 (1∶200, Cat. #sc-390351, Santa Cruz Biotechnology, Dallas, TX, USA), OSX (1∶500, Cat. #ab209484, Abcam, Cambridge, UK), OPG (1∶500, Cat. #DF6824, Affinity, Cincinnati, OH, USA), NFATc1 (1∶200, Cat. #sc-7294, Santa Cruz Biotechnology), c-Fos (1∶500, Cat. #66590-1-Ig, Proteintech, Rosemont, IL, USA), CTSK (1∶200, Cat. #sc-48353, Santa Cruz Biotechnology), and RANKL (1∶500, Cat. #bs-20647R, Bioss Antibodies, Woburn, MA, USA). Then, the sections were incubated with HRP-conjugated secondary antibodies at room temperature for one hour, followed by counterstaining with hematoxylin and sealing with neutral resin. Immunoreactivity was detected using an optical microscope (BX53, Olympus, Tokyo, Japan).
Western blotting analysis
For the extraction of protein from bone tissues, the femur was frozen in liquid nitrogen for 15 to 30 min, and then crushed and homogenized in RIPA lysis buffer (Cat. #C1053, APPLYGEN, Beijing, China). The homogenate was then centrifuged at 12000g and 4 ℃ for 15 min. The supernatant was collected, and protein concentrations were determined with the BCA protein assay (Pierce Chemical, Dallas, TX, USA) based on albumin standard. The proteins were subjected to SDS-PAGE, and then transferred onto a PVDF membrane. The membrane was blocked with 5% BSA and incubated overnight at 4 ℃ with specific primary antibodies as follows: β-catenin (1∶1000, Cat. #PK02151, Abmart), RUNX2 (1∶500, Cat. #sc-390351, Santa Cruz Biotechnology), OSX (1∶1000, Cat. #ab209484, Abcam), OPG (1∶1000, Cat. #DF6824, Affinity Biosciences), NFATc1 (1∶500, Cat. #sc-7294, Santa Cruz Biotechnology), c-Fos (1∶1000, Cat. #66590-1-Ig, Proteintech), CTSK (1∶500, Cat. #sc-48353, Santa Cruz Biotechnology), RANKL (1∶1000, Cat. #bs-20647R, Bioss Antibodies), DEC1 (1∶500, Cat. #sc-101023, Santa Cruz Biotechnology), Phospho-NF-κB p65 (Ser536) (1∶1000, Cat. #3033, Cell Signaling Technology), NF-κB p65 (1∶1000, Cat. #8242, Cell Signaling Technology), Phospho-IκBα (Ser32) (1∶1000, Cat. #2859, Cell Signaling Technology), IκBα (1∶1000, Cat. #9242, Cell Signaling Technology), matrix metallopeptidase 9 (MMP9; 1∶1000, Cat. #10375-2-AP, Proteintech), and β-actin (1∶1000, Cat. #AP0060, Bioworld Technology, Nanjing, China). After washing with TBST, the PVDF membrane was incubated with an HRP-conjugated secondary antibody at room temperature for one hour. Following the manufacturer's instructions, protein bands were detected using the ECL Western blotting detection system (Cat. #E423-01/02, Vazyme, Nanjing, China). Chemiluminescent signals were captured using image analysis software (ImageJ, 1.54i, NIH). β-Actin was used as an internal control.
Enzyme-linked immunosorbent assay (ELISA)
Mouse serum was collected from extracted eyeball blood. Blood samples were allowed to clot at room temperature for 30 min and then stored at 4 ℃ for an additional 30 min. The samples were centrifuged at 3000g and 4 ℃ for 10 min to obtain the serum. The concentrations of estradiol (Cat. #AF2566-A, AiFang Biological, Changsha, China), CTX-1 (Cat. #AF2808-A, AiFang Biological), IL-1β (Cat. #AF2040-A, AiFang Biological), IL-6 (Cat. #AF2163-A, AiFang Biological), and TNF-α (Cat. #KE10002, Proteintech) in the serum were determined by using an ELISA kit.
RNA isolation and real-time reverse transcription-PCR (RT-qPCR)
Total RNA was isolated from cultured cells using TRIzol reagent (Cat. #15596018CN, Thermo Fisher Scientific, Waltham, MA, USA), followed by cDNA synthesis using a reverse transcription kit (Cat. #R222-01, Vazyme). RT-qPCR was performed using the FastStar Universal SYBR Green Master (Cat. #Q311-02/03, Vazyme) and the 7300 real-time PCR system (Applied Biosystems, Foster City, CA, USA). β-Actin was used as an internal control for cDNA. Primer sequences are listed in Table 1.
Statistical analysis
All the data were expressed as mean ± standard deviation. Statistical analysis for multiple groups was performed using GraphPad Prism (GraphPad Prism, version 9, San Diego, CA, USA). Student's t-test or two-way ANOVA followed by Tukey's post hoc tests were performed to compare the differences between two groups or among more than two groups. Statistical significance was considered at a P-value of less than 0.05 (P < 0.05).
Results
DEC1 deficiency primarily reduced trabecular bone loss caused by OVX in mice
Dec1 knockout C57BL/6 mice (Dec1−/−) and littermate wild-type mice (Dec1+/+) were used to establish the OVX mice whose pathological features were similar to PMOP[5]. Estradiol is one of the most common and crucial estrogens. As shown in Supplementary Fig. 1, levels of the serum estradiol were significantly reduced in both Dec1+/+ and Dec1−/− mice three months post bilateral OVX operation, compared with those in the corresponding sham mice (Supplementary Fig. 1B). The weight was significantly increased in both Dec1+/+ and Dec1−/− OVX mice, compared with that in the corresponding sham mice (Supplementary Fig. 1C). The size of the uterus in Dec1+/+ OVX mice, rather than in Dec1−/− OVX mice, significantly decreased compared with that in corresponding sham mice (Supplementary Fig. 2A [available online]). The reason for no significant change in the uterus between the OVX and sham of Dec1−/− mice might be that the ERα expression levels of the uterus were increased in Dec1−/− OVX mice than in Dec1+/+ OVX mice (Supplementary Fig. 2B–2D). However, the expression levels of serum type I collagen C-terminal telopeptide (CTX-1), a byproduct of bone remodeling[27], were significantly increased in Dec1+/+ OVX mice than in Dec1+/+ sham mice, whereas the serum CTX-1 levels were not significantly different between Dec1−/− OVX mice and Dec1−/− sham mice (Supplementary Fig. 1D). The results suggested that the OVX model was established in both Dec1+/+ and Dec1−/− mice, and that bone turnover was increased in Dec1+/+ OVX mice but not in Dec1−/− OVX mice.
Next, we performed micro-CT scans on the femurs of the mice and conducted quantitative analyses of bone parameters to reveal differences in the microarchitecture of the bones among the four groups of mice (Dec1+/+ sham, Dec1+/+ OVX, Dec1−/− sham, and Dec1−/− OVX). As shown in Fig. 1, the bone mineral density of the trabecular bone of the femur was significantly lower in Dec1+/+ OVX mice than in Dec1+/+ sham mice, as indicated by the representative micro-CT reconstruction; however, this difference was not found between Dec1−/− OVX mice and Dec1−/− sham mice (Fig. 1A, left lane). Furthermore, there was no significant difference in cortical bone among the four groups (Fig. 1A, right lane). The results of the micro-CT quantitative analysis showed that in Dec1+/+ OVX mice, there was a significant decrease in BMD, BV/TV, Tb.N, and Tb.Th, but a slight increase in Tb.Sp, compared with that in Dec1+/+ sham mice. However, there were no significant differences in these bone mass parameters between Dec1−/− OVX and Dec1−/− sham mice (Fig. 1B–1F). Moreover, the bone mass parameters, including BMD, BV/TV, Tb.N, and Tb.Sp, in Dec1−/− OVX mice were significantly increased, compared with those in the Dec1+/+ OVX group (Fig. 1B–1D and Fig. 1F). Notably, there was no significant difference in the cortical bone of the femur, including Tt.Ar, Ct.Ar, and Ct.Th, among the four groups (Fig. 1G–1I). These changes were further demonstrated by H&E staining (Fig. 1J). These results implied that DEC1 deficiency primarily reduced trabecular bone loss caused by OVX in mice.
Figure
1.
DEC1 deficiency primarily reduced trabecular bone loss caused by OVX in mice.
A: Representative images of three-dimensional reconstruction of mouse femoral trabecular and cortical bone. B: The mean bone mineral density (BMD) value within each group (n = 6–8 in each group). C–I: Micro-CT analysis of bone volume/trabecular bone volume ratio (BV/TV) (C), trabecular bone number (Tb.N) (D), trabecular bone thickness (Tb.Th) (E), trabecular spacing (Tb.Sp) (F), total area (Tt.Ar) (G), cortical area (Ct.Ar) (H), and cortical thickness (Ct.Th) (I) of the distal femur metaphysis (n = 5–12 in each group). J: Representative images of H&E staining of the femur in each group of mice (n = 5–12 in each group). Data are presented as mean ± standard deviation and analyzed using two-way ANOVA followed by Tukey's honestly significant difference tests. *P < 0.05, **P < 0.01, ***P < 0.001, and nsP > 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; OVX, ovariectomy.
DEC1 deficiency mitigated the diminished osteogenic activity induced by OVX in mice
To determine the effect of Dec1 knockout on osteogenesis in mice, we examined the key proteins related to the differentiation and activity of osteoblasts, including β-catenin, RUNX2, OSX, and OPG, in the femur of the two genotypes of OVX mice by Western blotting. As shown in Fig. 2A and 2B, the expression levels of β-catenin, RUNX2, and OSX were significantly lower in Dec1+/+ OVX mice than in Dec1+/+ sham mice, while these proteins showed no significant difference between Dec1−/− OVX and Dec1−/− sham mice. Consistently, the IHC staining results showed that β-catenin, RUNX2, and OSX exhibited similar alterations in osteoblasts at the edge of trabeculae (Fig. 2C). Meanwhile, as a secretory protein, OPG was uniformly distributed in the growth plate. The OPG expression levels in the femoral growth plate were increased in Dec1−/− OVX mice, compared with those in Dec1+/+ OVX mice (Fig. 2C). These results indicated that Dec1 knockout alleviated the diminished osteogenic activity induced by OVX in mice.
Figure
2.
DEC1 deficiency mitigated the diminished osteogenic activity induced by OVX in mice.
A and B: The protein levels of β-catenin, RUNX2, OSX, and OPG in the femur in the two genotypes of OVX mice by Western blotting analysis (n = 5 in each group). C: The protein levels of β-catenin, RUNX2, OSX, and OPG in the femur of the four groups of mice by immunostaining analysis. Scale bar, 20 μm or 50 μm (n = 5 in each group). Red arrowheads indicate the positive stainings in immunohistochemistry. Data are presented as mean ± standard deviation and analyzed using two-way ANOVA followed by Tukey's honestly significant difference tests. *P < 0.05, **P < 0.01, ****P < 0.0001, and nsP > 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; OVX, ovariectomy; RUNX2, runt-related transcription factor 2; OSX, osterix; OPG, osteoprotegerin.
DEC1 deficiency alleviated the enhanced bone resorption induced by OVX in mice
We analyzed the expression of osteoclast-specific genes to elucidate the effect of DEC1 on the formation and function of osteoclasts induced by OVX. The Western blotting results showed that the expression levels of RANKL were significantly increased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice, whereas it did not significantly increase in Dec1−/− OVX mice, compared with those in Dec1−/− sham mice (Fig. 3A). Furthermore, the results of IHC staining showed that RANKL was secreted into the growth plate (Fig. 3C). Combined with the fourth line of Fig. 2A and the fourth line of Fig. 3A, we observed that the RANKL/OPG ratio was significantly increased in Dec1+/+ OVX mice, compared with that in Dec1+/+ sham mice, but not in Dec1−/− OVX mice, compared with that in Dec1−/− sham mice (Fig. 3B). In addition, the protein levels of NFATc1, c-Fos, and CTSK were significantly increased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice, while these indicators did not increase in Dec1−/− OVX mice, compared with those in Dec1−/− sham mice (Fig. 3A and 3B). IHC results revealed consistent protein levels of NFATc1, c-Fos, and CTSK in osteoclasts at the edge of trabeculae (Fig. 3C). These results demonstrated that Dec1 knockout alleviated the enhanced bone resorption induced by OVX in mice.
Figure
3.
DEC1 deficiency alleviated the enhanced bone resorption induced by OVX in mice.
A and B: The protein levels of NFATc1, c-Fos, CTSK, and RANKL in the femur in the two types of OVX mice by Western blotting analysis (n = 5 in each group). C: The protein levels of NFATc1, c-Fos, CTSK, and RANKL in the femur in the two types of OVX mice by immunostaining analysis. Scale bar, 20 μm or 50 μm. Red arrowheads indicate the positive stainings in immunohistochemistry. Data are presented as mean ± standard deviation and analyzed using two-way ANOVA followed by Tukey's honestly significant difference tests. **P < 0.01, ***P < 0.001, and nsP > 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; OVX, ovariectomy; NFATc1, nuclear factor of activated T cells 1; CTSK, cathepsin K; RANKL, receptor activator of nuclear factor kappa-B ligand.
DEC1 deficiency mitigated the increased inflammation induced by OVX in mice
OVX in mice leads to decreased estrogen levels, resulting in an increase in systemic inflammatory cytokines[28]. Inflammatory cytokines, such as IL-1β and TNF-α, have long been implicated in osteoblastic bone loss. These cytokines promote the production of RANKL by both osteoblast precursor cells and mature osteoblasts, and also reduce OPG production[29]. To investigate the role of DEC1 in the inflammatory process induced by OVX, we collected the serum from the orbital blood of model mice for ELISA detection. As shown in Fig. 4, the serum levels of IL-1β and IL-6, but not TNF-α, were significantly increased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice. Although IL-1β levels were increased in both Dec1+/+ and Dec1−/− OVX mice, compared with those in corresponding sham mice, the serum levels of IL-1β and IL-6, but not TNF-α, significantly decreased in Dec1−/− OVX mice, compared with those in Dec1+/+ OVX mice. These data indicated that Dec1 deletion inhibited the increased inflammation induced by OVX.
Figure
4.
DEC1 deficiency mitigated the increased inflammation induced by OVX in mice.
A–C: The serum amounts of IL-1β, IL-6, and TNF-α by enzyme-linked immunosorbent assay in the two genotypes of OVX mice (n = 6–14 in each group). Data are presented as mean ± standard deviation and analyzed using two-way ANOVA followed by Tukey's honestly significant difference tests. *P < 0.05, **P < 0.01, ****P < 0.0001, and nsP > 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; OVX, ovariectomy; ILs, interleukins; TNF-α, tumor necrosis factor-α.
Decreased DEC1 in BMSCs and increased DEC1 in BMMs from Dec1+/+ OVX mice
BMSCs are pluripotent stem cells with self-renewal and multidirectional differentiation capabilities. As a vital source of osteoprogenitor cells, BMSCs are crucial for maintaining normal physiological functions of bone tissues, and promoting bone repair and regeneration[30]. BMMs are a type of mononuclear cells with the potential to differentiate into various cell types[31]. When stimulated by factors such as RANKL and M-CSF, BMMs may differentiate into osteoclasts[32].
The heterogeneity of DEC1 expression in cells has significant implications for various biological processes. To investigate the DEC1 differential expression in BMSCs and BMMs, we isolated and cultured the two types of cells from Dec1+/+ OVX or sham mice, respectively (Fig. 5A). We found that the protein levels of DEC1 were significantly decreased in BMSCs but increased in BMMs from Dec1+/+ OVX mice, compared with those from Dec1+/+ sham mice (Fig. 5B–5E). These results indicated that the effects of OVX on DEC1 expression in BMSCs and BMMs of Dec1+/+ mice were opposite.
Figure
5.
The DEC1 expression was decreased in BMSCs but increased in BMMs from the wild-genotype OVX mice compared with that of the sham mice.
A: The flow chart of isolating BMSCs and BMMs from the wild-type OVX mice (n = 5 in each group). B and C: DEC1 expression in BMSCs in Dec1+/+ mice by Western blotting analysis (n = 5 in each group). D and E: DEC1 expression in BMMs in Dec1+/+ mice by Western blotting analysis (n = 5 in each group). Data are presented as mean ± standard deviation and analyzed using Student's t-test. *P < 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; BMSCs, bone marrow mesenchymal stem cells; BMMs, bone marrow-derived macrophages; OVX, ovariectomy.
DEC1 deficiency enhanced osteoblast activity along with attenuating the NF-κB pathway and increasing the RANKL/OPG ratio in BMSCs from the OVX mice
To determine the effect of DEC1 in BMSCs and BMMs on bone loss induced by OVX, we isolated and cultured BMSCs and BMMs from both Dec1+/+ and Dec1−/− OVX or sham mice and analyzed their osteogenic differentiation capabilities (Fig. 6A). The results showed that the mRNA levels of genes Runx2 and Osx, which contributed to bone differentiation, as well as the gene alkaline phosphatase (Alp) that promoted bone activity, were significantly decreased in Dec1+/+ OVX BMSCs, compared with those in Dec1+/+ sham BMSCs, whereas they did not decrease in Dec1−/− OVX BMSCs, compared with those in Dec1−/− sham BMSCs (Fig. 6B–6E). Moreover, the phosphorylation levels of IκBα and P65 were significantly increased in Dec1+/+ OVX BMSCs, compared with those in Dec1+/+ sham BMSCs, but did not increase in Dec1−/− OVX BMSCs, compared with those in Dec1−/− sham BMSCs (Fig. 6F and 6G). Notably, the phosphorylation levels of IκBα and P65 were significantly decreased in Dec1−/− OVX BMSCs, compared with those in Dec1+/+ OVX BMSCs, which was consistent with the result of RANKL/OPG in vivo (Fig. 3B). Furthermore, the Rankl/Opg ratio was significantly increased in both Dec1+/+ and Dec1−/− OVX BMSCs, compared with those in corresponding sham BMSCs, and the increased Rankl/Opg ratio was significantly lower in Dec1−/− OVX BMSCs than in Dec1+/+ OVX BMSCs (Fig. 6H). These results indicated that DEC1 deficiency enhanced osteoblast activity while attenuating the NF-κB pathway and increasing the RANKL/OPG ratio in BMSCs from the OVX mice.
Figure
6.
DEC1 deficiency attenuated the NF-κB activation but increased osteoblast activity and the Rankl/Opg ratio in BMSCs from the two genotypes of the OVX mice.
A: The flow chart of isolating BMSCs and BMMs from the two genotypes of the OVX mice (n = 5 in each group). B–E: The mRNA levels of Runx2, Osx, Alp, and Ocn in BMSCs from the two genotypes of the mice (n = 3 in each group). F and G: The protein levels of p-P65, P65, p-IκBα, and p-IκBα in BMSCs from the two genotypes of the mice by Western blotting (n = 5 in each group). H: The mRNA levels of Rankl/Opg in BMSCs from the two genotypes of the mice (n = 3 in each group). Data are presented as mean ± standard deviation and analyzed using two-way ANOVA followed by Tukey's honestly significant difference tests. *P < 0.05, **P < 0.01, ***P < 0.001, and nsP > 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; RANKL, receptor activator of nuclear factor kappa-B ligand; OPG, osteoprotegerin; OVX, ovariectomy.
DEC1 deficiency inhibited osteoclast function along with repressing the NF-κB signaling and reducing pro-inflammatory cytokine production in BMMs from the OVX mice
We further determined the effects of DEC1 in BMMs on the bone loss induced by OVX. We found that the mRNA levels of the genes Nfatc1 and c-Fos, which promoted osteoclast differentiation, Ctsk and Trap, which were linked to osteoclast activity, were significantly increased in Dec1+/+ OVX BMMs, compared with those in Dec1+/+ sham BMMs; however, the upregulation of these genes was not observed in Dec1−/− OVX BMMs, compared with those in Dec1−/− sham BMMs (Fig. 7A–7D). Furthermore, the expression levels of these genes were significantly lower in Dec1−/− OVX BMMs than in Dec1+/+ OVX BMMs (Fig. 7A–7D). Consistent with the mRNA levels, the protein levels of NFATc1 showed a similar trend and pattern (Fig. 7E and Fig. 7F). Although the protein levels of MMP9, which was also linked to osteoclast activity, were significantly increased in both Dec1+/+ and Dec1−/− OVX BMMs, compared with those in the corresponding sham BMMs, the increased protein levels of MMP9 were significantly lower in Dec1−/− OVX BMMs than in Dec1+/+ sham BMMs (Fig. 7E and 7F).
Figure
7.
DEC1 deficiency attenuated the osteoclast activation along with decreasing the NF-κB activation and reducing the release of pro-inflammatory cytokines in BMMs from the two genotypes of the OVX mice.
A–D: The mRNA levels of Nfatc1, c-Fos, Ctsk, and Trap in BMMs from the four groups of mice (n = 3 in each group). E and F: The protein levels of NFATc1, MMP9, p-P65/P65, and p-IκBα/IκBα in BMMs from the four groups of mice (n = 5 in each group). G–J: The mRNA levels of Il1b, Il6, Tnfa, and Ifng in BMMs from the four groups of mice (n = 3 in each group). Data are presented as mean ± standard deviation and analyzed using two-way ANOVA followed by Tukey's honestly significant difference tests. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and nsP > 0.05. Abbreviations: DEC1, differentiated embryo-chondrocyte expressed gene 1; BMMs, bone marrow-derived macrophages; OVX, ovariectomy; NFATc1, nuclear factor of activated T cells 1; MMP9, matrix metallopeptidase 9.
BMMs are precursor cells to various cell types, and are capable of not only differentiating into osteoclasts but also into macrophages that are crucial components of the immune system and are capable of engulfing pathogens, dead cells, and other debris[33]. During the process of bone remodeling, BMMs influence bone formation and resorption through the release of various cytokines, including growth factors, and pro-inflammatory cytokines[34–35]. To determine whether the reduced osteoclast function in Dec1−/− OVX BMMs was correlated with inflammation, we measured the mRNA levels of pro-inflammatory cytokines, such as Il1b, Il6, Tnfa, and Ifng, in BMMs from the two genotypes of the OVX mice. As a result, the mRNA levels of Il1b, Il6, Tnfa, and Ifng were significantly increased in Dec1+/+ OVX BMMs, compared with those in Dec1+/+ sham BMMs, but not in Dec1−/− OVX BMMs, compared with those in Dec1−/− sham BMMs (Fig. 7G–7J). Moreover, the mRNA levels of these inflammatory cytokines were significantly lower in Dec1−/− OVX BMMs than in Dec1+/+ OVX BMMs. Furthermore, the phosphorylation levels of IκBα and P65 were significantly increased in Dec1+/+ OVX BMMs, compared with those in Dec1+/+ sham BMMs, but did not increase in Dec1−/− OVX BMMs, compared with those in Dec1−/− sham BMMs (Fig. 7K and 7L). Notably, the phosphorylation levels of IκBα and P65 were significantly lower in Dec1−/− OVX BMMs than in Dec1+/+ OVX BMMs. These results indicated that the DEC1 deficiency might inhibit osteoclast function while repressing the NF-κB signaling and reducing pro-inflammatory cytokine production in BMMs from the OVX mice.
Discussion
PMOP is a condition affecting postmenopausal women characterized by a decrease in bone mass because of estrogen deficiency. The underlying mechanism involves the direct effects of estrogen on osteoblasts and osteoclasts, which in turn affects bone homeostasis. Furthermore, estrogen deficiency leads to an increase in inflammatory factors in the body, significantly contributing to PMOP[28]. Studies have reported that DEC1 may be associated with immune response[22]. The immune checkpoint therapy upregulated the transcription factor DEC1 in tumor antigen-specific CD8+ and CD4+ T cells, while DEC1 deficiency inhibited immune checkpoint therapy-induced macrophage transformation from M2 to M1 subtype[36]. In addition, DEC1 overexpression increased the expression of pro-inflammatory cytokines, such as TNFα, IL-1β, and IL-6[37]. Therefore, DEC1 may play a pivotal role in intracellular signaling and the regulation of inflammatory responses.
The current study provided some evidence both in vivo and in vitro to support that DEC1 deficiency protected against bone loss induced by inflammation. Evidence in vivo: (1) The phenotypes of bone loss significantly increased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice, but no difference was observed between Dec1−/− OVX and Dec1−/− sham mice. Likewise, compared with wild-type mice, DEC1 deficiency reduced the phenotypes of bone loss induced by OVX in mice. (2) The osteogenesis-related proteins, such as β-catenin and RUNX2, significantly decreased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice, but no difference was found between Dec1−/− OVX and Dec1−/− sham mice. Similarly, DEC1 deficiency inhibited the decreased osteogenesis induced by OVX in the mice. (3) The bone destruction-related proteins, such as NFATc1, c-Fos, and CTSK, as well as the RANKL/OPG ratio, were significantly increased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice, but no difference was observed between Dec1−/− OVX and Dec1−/− sham mice. In the same way, DEC1 deficiency decreased the increased bone destruction induced by OVX in the mice. (4) The serum levels of IL-1β and IL-6 were significantly lower in Dec1−/− OVX mice than those in Dec1+/+ OVX mice, implying that DEC1 deficiency inhibited the OVX-induced serum levels of IL-1β and IL-6. In conclusion, osteogenic capacity was protected, whereas osteoclastogenic capacity was inhibited in Dec1−/− OVX mice. Likewise, the release of pro-inflammatory cytokines was significantly inhibited in Dec1−/− OVX mice.
The OVX mice model mimics PMOP through the combined effects of multiple cell types[5]. Evidence in vitro: (1) DEC1 expression levels were decreased in BMSCs but increased in BMMs from OVX-induced Dec1+/+ mice. (2) Consistent with the in vivo results, DEC1 deficiency increased the expression levels of osteogenesis-related genes, such as Runx2, Osx, Alp, and Ocn, but decreased the Rankl/Opg ratio that was related to osteoclasts in BMSCs induced by OVX. Whereas DEC1 deficiency decreased the osteoclast-related genes, such as Nfatc1, c-Fos, Ctsk, and Trap, along with the decreased expression of pro-inflammatory cytokines, such as Il1b, Il6, Tnfa, and Ifng, in BMMs induced by OVX, the differential expression of DEC1 in BMSCs and BMMs of Dec1+/+ OVX mice prompted us to further consider the following. In vitro experiments demonstrated that OVX caused the decreased expression levels of DEC1 in Dec1+/+ BMSCs that had the potential for osteogenic differentiation, downregulating the osteogenic capacity. This finding was consistent with our previous study[25]. In contrast, OVX induced the increased expression of DEC1 in Dec1+/+ BMMs that possessed potential for osteoclast differentiation, leading to the increased secretion of pro-inflammatory cytokines, such as IL-1β and IL-6. The increased pro-inflammatory cytokines have been found to inhibit osteogenesis on one hand[38–39], and promote osteoclastogenesis on the other hand[39–40]. Conversely, Dec1 knockout resulted in a significant inhibition of pro-inflammatory cytokine secretion, thereby rescuing the imbalance in bone remodeling in OVX mice. Pro-inflammatory cytokines are released into the bone microenvironment, contributing to the development of osteoblasts and osteoclasts, and in turn affecting bone metabolism and the development of osteoporosis[41–42]. The NF-κB pathway serves as a critical pathway for pro-inflammatory cytokines (such as IL-1β, IL-6, and TNF-α)[43]. The phosphorylation of IκBα activates NF-κB that mediates cytokine release[44–45]. To investigate the role of DEC1 in bone loss induced by OVX, we detected the DEC1 expression in both BMSCs and BMMs from Dec1+/+ OVX mice and the phosphorylation levels of IκBα and P65 in both BMSCs and BMMs from the two genotypes (Dec1+/+ and Dec1−/−) of OVX mice. As a result, we found that the phosphorylation levels of IκBα and P65 were significantly increased in Dec1+/+ OVX mice, compared with those in Dec1+/+ sham mice, but DEC1 deficiency did not elevate the phosphorylation levels of IκBα and P65 in either BMSCs or BMMs induced by OVX, implying that the upregulation of pro-inflammatory cytokines, such as IL-1β, IL-6, and TNF-α induced by OVX was mediated by the increased activation of the NF-κB pathway. DEC1 deficiency inhibited the NF-κB pathway induced by OVX, thereby reducing cytokine levels. Consequently, the reduced cytokines attenuated the decrease of osteogenesis and the increase of osteoclastogenesis induced by OVX. These results were further supported by the findings that the IKK/NF-κB pathway regulated the osteogenesis and osteoclastogenesis in vivo and in vitro[46].
However, there are certain limitations to the current study. First, we used global rather than conditional knockout mice to construct the OVX model. Therefore, we isolated BMSCs and BMMs from Dec1−/− OVX mice and Dec1+/+ OVX mice, respectively, to determine the differential role of DEC1 in osteoblasts and osteoclasts. Second, the effects of DEC1 on the NF-κB pathway in BMSCs and BMMs remain to be demonstrated in future studies.
In summary, DEC1 deficiency inhibits the NF-κB pathway induced by OVX, thereby reducing the levels of cytokines, such as IL-1β and IL-6, and subsequently inhibits the decrease of osteogenesis and the increase of osteoclastogenesis caused by OVX. Taken together, DEC1 deficiency protects against bone loss induced by OVX by inhibiting inflammation. These findings provide a novel understanding of postmenopausal osteoporosis development and may offer potential avenues for developing disease intervention strategies.
Rodríguez-Bermúdez G, García Laencina PJ. Analysis of EEG signals using nonlinear dynamics and chaos: a review[J]. Appl Math Inf Sci, 2015, 9(5): 2309–2321.
[3]
Tzallas AT, Tsipouras MG, Fotiadis DI. Automatic seizure detection based on time-frequency analysis and artificial neural networks[J]. Comput Intell Neurosci, 2007, 2007: 80510.
[4]
Pachori BB, Patidar S. Epileptic seizure classification in EEG signals using second-order difference plot of intrinsic mode functions[J]. Comput Meth Progr Biomed, 2014, 113(2): 494–502. doi: 10.1016/j.cmpb.2013.11.014
[5]
Niknazar M, Mousavi SR, Vahdat BV, et al. A new framework based on recurrence quantification analysis for epileptic seizure detection[J]. IEEE J Biomed Health Inf, 2013, 17(3): 572–578. doi: 10.1109/JBHI.2013.2255132
[6]
Guo L, Rivero D, Pazos A. Epileptic seizure detection using multiwavelet transform based approximate entropy and artificial neural networks[J]. J Neurosci Meth, 2010, 193(1): 156–163. doi: 10.1016/j.jneumeth.2010.08.030
[7]
Sharmila A, Geethanjali P. DWT based detection of epileptic seizure from EEG signals using naive Bayes and k-NN classifiers[J]. IEEE Access, 2016, 4: 7716–7727. doi: 10.1109/ACCESS.2016.2585661
[8]
Faust O, Bairy MG. Nonlinear analysis of physiological signals: a review[J]. J Mech Med Biol, 2012, 12(4): 1240015. doi: 10.1142/S0219519412400155
[9]
Sanei S, Chambers JA. EEG signal processing[M]. Hoboken: John Wiley & Sons, Ltd, 2007.
[10]
Gotman J, Ives JR, Gloor P. Automatic recognition of inter-ictal epileptic activity in prolonged EEG recordings[J]. Electroencephalogr Clin Neurophysiol, 1979, 46(5): 510–520. doi: 10.1016/0013-4694(79)90004-X
[11]
Ayoubian L, Lacoma H, Gotman J. Automatic seizure detection in SEEG using high frequency activities in wavelet domain[J]. Med Eng Phys, 2013, 35(3): 319–328. doi: 10.1016/j.medengphy.2012.05.005
[12]
Andrzejak RG, Lehnertz K, Mormann F, et al. Indications of nonlinear deterministic and finite-dimensional structures in time series of brain electrical activity: dependence on recording region and brain state[J]. Phys Rev E, 2001, 64(6): 061907. doi: 10.1103/PhysRevE.64.061907
[13]
Temko A, Thomas E, Marnane W, et al. EEG- based neonatal seizure detection with support vector machines[J]. Clin Neurophysiol, 2011, 122(3): 464–473. doi: 10.1016/j.clinph.2010.06.034
[14]
Liu YX, Zhou WD, Yuan Q, et al. Au- tomatic seizure detection using wavelet transform and SVM in long- term intracranial EEG[J]. IEEE Trans Neural Syst Rehabil Eng, 2012, 20(6): 749–755. doi: 10.1109/TNSRE.2012.2206054
[15]
Bhattacharyya A, Pachori RB, Upadhyay A, et al. Tunable-q wavelet transform based multiscale entropy measure for automated classification of epileptic EEG signals[J]. Appl Sci, 2017, 7(4): 385. doi: 10.3390/app7040385
[16]
Acharya UR, Sree SV, Chattopadhyay S, et al. Application of recurrence quantification analysis for the automated identification of epileptic EEG signals[J]. Int J Neural Syst, 2011, 21(3): 199–211. doi: 10.1142/S0129065711002808
[17]
Shoeb AH. Application of machine learning to epileptic seizure onset detection and treatment[D]. Cambridge: Massachusetts Institute of Technology, 2009.
[18]
Gandhi T, Panigrahi BK, Anand S. A comparative study of wavelet families for EEG signal classification[J]. Neurocomputing, 2011, 74(17): 3051–3057. doi: 10.1016/j.neucom.2011.04.029
[19]
Swami P, Gandhi TK, Panigrahi BK, et al. A novel robust diagnostic model to detect seizures in electroencephalography[J]. Expert Syst Appl, 2016, 56: 116–130. doi: 10.1016/j.eswa.2016.02.040
[20]
Nicolaou N, Georgiou J. Detection of epileptic electroencephalogram based on permutation entropy and support vector machines[J]. Expert Syst Appl, 2012, 39(1): 202–209. doi: 10.1016/j.eswa.2011.07.008
[21]
Chua KC, Chandran V, Acharya UR, et al. Automatic identification of epileptic electroencephalography signals using higher-order spectra[J]. Proc Inst Mech Eng, Part H: J Eng Med, 2009, 223(4): 485–495. doi: 10.1243/09544119JEIM484
[22]
Chua KC, Chandran V, Acharya UR, et al. Application of higher order spectra to identify epileptic EEG[J]. J Med Syst, 2011, 35(6): 1563–1571. doi: 10.1007/s10916-010-9433-z
[23]
Faust O, Acharya UR, Min LC, et al. Automatic identification of epileptic and background EEG signals using frequency domain parameters[J]. Int J Neural Syst, 2010, 20(2): 159–176. doi: 10.1142/S0129065710002334
[24]
Ullah I, Hussain M, Qazi EUH, et al. An automated system for epilepsy detection using EEG brain signals based on deep learning approach[J]. Expert Syst Appl, 2018, 107: 61–71. doi: 10.1016/j.eswa.2018.04.021
[25]
Acharya UR, Oh SL, Hagiwara Y, et al. Deep convolutional neural network for the automated detection and diagnosis of seizure using EEG signals[J]. Comput Biol Med, 2018, 100: 270–278. doi: 10.1016/j.compbiomed.2017.09.017
[26]
Guo L, Rivero D, Dorado J, et al. Automatic feature extraction using genetic programming: an application to epileptic EEG classification[J]. Expert Syst Appl, 2011, 38(8): 10425–10436. doi: 10.1016/j.eswa.2011.02.118
[27]
Tzallas AT, Tsipouras MG, Fotiadis DI, et al. Epileptic seizure detection in EEGs using time-frequency analysis[J]. IEEE Trans Inf Technol Biomed, 2009, 13(5): 703–710. doi: 10.1109/TITB.2009.2017939
[28]
Guo L, Rivero D, Dorado J, et al. Automatic epileptic seizure detection in EEGs based on line length feature and artificial neural networks[J]. J Neurosci Meth, 2010, 191(1): 101–109. doi: 10.1016/j.jneumeth.2010.05.020
[29]
Kumar Y, Dewal ML, Anand RS. Epileptic seizures detection in EEG using DWT-based ApEn and artificial neural network[J]. Signal, Image Video Process, 2014, 8(7): 1323–1334. doi: 10.1007/s11760-012-0362-9
[30]
Sharma R, Pachori RB. Classification of epileptic seizures in EEG signals based on phase space representation of intrinsic mode functions[J]. Expert Syst Appl, 2015, 42(3): 1106–1117. doi: 10.1016/j.eswa.2014.08.030
[31]
Sharma M, Pachori RB, Acharya UR. A new approach to characterize epileptic seizures using analytic time-frequency flexible wavelet transform and fractal dimension[J]. Pattern Recognit Lett, 2017, 94: 172–179. doi: 10.1016/j.patrec.2017.03.023
[32]
Acharya UR, Sree SV, Suri JS. Automatic detection of epileptic EEG signals using higher order cumulant features[J]. Int J Neural Syst, 2011, 21(5): 403–414. doi: 10.1142/S0129065711002912
[33]
Subasi A, Gursoy MI. EEG signal classification using PCA, ICA, LDA and support vector machines[J]. Expert Syst Appl, 2010, 37(12): 8659–8666. doi: 10.1016/j.eswa.2010.06.065
[34]
Acharya UR, Chua CK, Lim TC, et al. Automatic identification of epileptic EEG signals using nonlinear parameters[J]. J Mech Med Biol, 2009, 9(4): 539–553. doi: 10.1142/S0219519409003152
[35]
Djemal R, Bazyed AG, Belwafi K, et al. Three-class EEG-based motor imagery classification using phase-space reconstruction technique[J]. Brain Sci, 2016, 6(3): 36. doi: 10.3390/brainsci6030036
[36]
Bhattacharyya A, Singh L, Pachori RB. Identification of epileptic seizures from scalp EEG signals based on TQWT[M]//Tanveer M, Pachori RB. Machine Intelligence and Signal Analysis. Singapore: Springer, 2019: 209–221.
[37]
Takens F. Detecting strange attractors in turbulence[M]//Rand D, Young LS. Dynamical Systems and Turbulence, Warwick 1980. Berlin, Heidelberg: Springer, 1981: 366–381.
[38]
Sze V, Chen YH, Yang TJ, et al. Efficient processing of deep neural networks: a tutorial and survey[J]. Proc IEEE, 2017, 105(12): 2295–2329. doi: 10.1109/JPROC.2017.2761740
[39]
Kermany DS, Goldbaum M, Cai WJ, et al. Identifying medical diagnoses and treatable diseases by image-based deep learning[J]. Cell, 2018, 172(5): 1122–1131. doi: 10.1016/j.cell.2018.02.010
[40]
Oquab M, Bottou L, Laptev I, et al. Learning and transferring mid-level image representations using convolutional neural networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Columbus: IEEE, 2014: 1717–1724.
[41]
Rawat W, Wang ZH. Deep convolutional neural networks for image classification: a comprehensive review[J]. Neural Comput, 2017, 29(9): 2352–2449. doi: 10.1162/neco_a_00990
Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems. Lake Tahoe: Curran Associates Inc., 2012: 1097–1105.
[44]
Kliková B, Raidl A. Reconstruction of phase space of dynamical systems using method of time delay[C]//Proceedings of Contributed Papers, Part III. Prague: MATFYZPRESS, 2011: 83–87.
[45]
Huffaker R. Phase space reconstruction from economic time series data: improving models of complex real-world dynamic systems[J]. Int J Food Syst Dyn, 2010, 1(3): 184–193.
[46]
Li HY, Bao SS, Xuan YQ. Parameter selection for phase space reconstruction in hydrological series and rationality analysis of its chaotic characteristics[J]. EPiC Ser Eng, 2018, 3: 1171–1183. doi: 10.29007/2fb8
Liu ZW, Chen G, Dong CF, et al. Intelligent assistant diagnosis for pediatric inguinal hernia based on a multilayer and unbalanced classification model. Front Physiol, 2023, 14: 1105891.
DOI:10.3389/fphys.2023.1105891
2.
Boubchir L. Editorial commentary on special issue of Advances in EEG Signal Processing and Machine Learning for Epileptic Seizure Detection and Prediction. J Biomed Res, 2020, 34(3): 149-150.
DOI:10.7555/JBR.34.20200700
Authors and Reviewers
DownLoad:
