JMIR Formative Research

View query in Help articles search

Field

e.g. mhealth

Operator

Field

Criteria

e.g. mhealth

Search Results (1 to 10 of 13 Results)

Download search results: CSV END BibTex RIS

Sort by:

Convolutional Neural Network Models for Visual Classification of Pressure Ulcer Stages: Cross-Sectional Study

Alex Net, a pioneering CNN in image classification, consists of 8 layers: 5 convolutional layers with varying filter numbers and 3 fully connected layers. It employs Re LU activations, max pooling, and dropout for regularization. The introduction of Re LU and dropout layers in Alex Net reduced training times and prevented overfitting, whereas its deep architecture allowed for the learning of complex features, enhancing classification accuracy [18].

Changbin Lei, Yan Jiang, Ke Xu, Shanshan Liu, Hua Cao, Cong Wang

JMIR Med Inform 2025;13:e62774

Download Citation: END BibTex RIS

Download PDF Download XML

The Application of Mask Region-Based Convolutional Neural Networks in the Detection of Nasal Septal Deviation Using Cone Beam Computed Tomography Images: Proof-of-Concept Study

To identify targets and detect deviated nasal septum, Mask R-CNN was used to segment the nasal septum at the pixel level. The Mask R-CNN architecture consisted of faster R-CNN, Region of Interest (Ro I) alignment (Ro IAlign) technique, and feature pyramid networks [20]. Mask R-CNN is an improved version of faster R-CNN, a model used for object detection, which includes a segment prediction branch for each Ro I, allowing it to perform both detection and pixel-level segmentation at the same time.

Shishir Shetty, Auwalu Saleh Mubarak, Leena R David, Mhd Omar Al Jouhari, Wael Talaat, Natheer Al-Rawi, Sausan AlKawas, Sunaina Shetty, Dilber Uzun Ozsahin

JMIR Form Res 2024;8:e57335

Download Citation: END BibTex RIS

Download PDF Download XML

Performance Test of a Well-Trained Model for Meningioma Segmentation in Health Care Centers: Secondary Analysis Based on Four Retrospective Multicenter Data Sets

Recent results have reported that well-trained convolutional neural network (CNN) models can show state-of-the-art performance for meningioma segmentation on MR imaging (MRI) with a Dice ratio of more than 0.900 [5-10]. On the one hand, these studies provided a series of CNN models for automated tumor detection and volumetric assessment, indicating convenient radiological tools to facilitate patient management in clinical practice.

Chaoyue Chen, Yuen Teng, Shuo Tan, Zizhou Wang, Lei Zhang, Jianguo Xu

J Med Internet Res 2023;25:e44119

Download Citation: END BibTex RIS

Download PDF Download XML

Predicting the Risk of Total Hip Replacement by Using A Deep Learning Algorithm on Plain Pelvic Radiographs: Diagnostic Study

CNN: convolutional neural network; ROI: region of interest. The images were initially labeled as THR or non-THR according to the surgical reports. For the THR group, we reviewed all surgical reports to ensure that patients had undergone THR and then acquired the corresponding images for training. For the non-THR group, we could only confirm that patients had not received THR within the subsequent 3 months at our hospital.

Chih-Chi Chen, Cheng-Ta Wu, Carl P C Chen, Chia-Ying Chung, Shann-Ching Chen, Mel S Lee, Chi-Tung Cheng, Chien-Hung Liao

JMIR Form Res 2023;7:e42788

Download Citation: END BibTex RIS

Download PDF Download XML

Chinese Clinical Named Entity Recognition From Electronic Medical Records Based on Multisemantic Features by Using Robustly Optimized Bidirectional Encoder Representation From Transformers Pretraining Approach Whole Word Masking and Convolutional Neural Networks: Model Development and Validation

To improve the ability to capture details and extract features of medical NER models, many studies added Word2 Vec with static representation [38], Global Vectors for Word Representation [39] with static representation, Embeddings from Language Models (ELMo) [40,41] with dynamic representation, CNN [42], and attention mechanism [43] to the Bi LSTM-CRF model.

Weijie Wang, Xiaoying Li, Huiling Ren, Dongping Gao, An Fang

JMIR Med Inform 2023;11:e44597

Download Citation: END BibTex RIS

Download PDF Download XML

Assessing the Generalizability of Deep Learning Models Trained on Standardized and Nonstandardized Images and Their Performance Against Teledermatologists: Retrospective Comparative Study

The same 3 CNN models were then additionally trained with a Res Net-18 backbone on either the ISIC 2019 (CNN nonstandardized [CNN-NS]) or Mole Map (CNN standardized [CNN-S] and CNN standardized number 2 [CNN-S2]) data sets. CNN-NS was trained on 25,331 nonstandardized ISIC dermoscopic images consisting of 8 skin conditions (Table 1). We define nonstandardized images as images that are taken using multiple image capture technologies (Figure 2).

Ayooluwatomiwa I Oloruntoba, Tine Vestergaard, Toan D Nguyen, Zhen Yu, Maithili Sashindranath, Brigid Betz-Stablein, H Peter Soyer, Zongyuan Ge, Victoria Mar

JMIR Dermatol 2022;5(3):e35150

Download Citation: END BibTex RIS

Download PDF Download XML

Predicting Abnormalities in Laboratory Values of Patients in the Intensive Care Unit Using Different Deep Learning Models: Comparative Study

In our case, we used a 1 D multiple CNN (M-CNN), where the kernels (filters) move along the time axis performing convolution operations on all features. The kernel size defines how many time steps 1 kernel covers at any point in time. Aside from the normal CNN that takes 1 input stream, we developed an architecture that takes 2 streams of the input sequences in parallel. Each stream will be processed with different filters.

Ahmad Ayad, Ahmed Hallawa, Arne Peine, Lukas Martin, Lejla Begic Fazlic, Guido Dartmann, Gernot Marx, Anke Schmeink

JMIR Med Inform 2022;10(8):e37658

Download Citation: END BibTex RIS

Download PDF Download XML

State-of-the-Art Deep Learning Methods on Electrocardiogram Data: Systematic Review

Almost the half of them (10/23, 43.5%) applied a CNN structure. Regarding the MIT-BIH Arrhythmia Database, the best accuracy (99.94%) was achieved by Wang et al [185], who introduced a fused autoencoder-CNN network to classify 6 different ECG rhythms. However, a high percentage of the studies that managed to classify data originating from the same database implemented a CNN structure.

Georgios Petmezas, Leandros Stefanopoulos, Vassilis Kilintzis, Andreas Tzavelis, John A Rogers, Aggelos K Katsaggelos, Nicos Maglaveras