What does this research mean for the field?

A multimodal model integrating clinical variables, radiomics features, and deep learning features accurately identifies sarcopenia and severe sarcopenia in patients with gastric cancer. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

To develop and validate a multimodal model combining clinical variables, radiomics, and deep learning features from computed tomography scans to detect sarcopenia and severe sarcopenia in patients with gastric cancer.

September 15, 2026Journal of Cachexia Sarcopenia and MuscleOpen Access

Integrated Assessment of Sarcopenia in Patients with Gastric Cancer Using Deep Learning and Radiomics

View Full Paper

Ask AI

Bookmark

Authors

HZHuaiqing Zhi JZJingwei Zheng HCHao Chen

Discussion

Member takes

Overview

Multicenter cohort study demonstrates accurate identification of sarcopenia in gastric cancer using multimodal radiomics and deep learning, indicating potential for automated clinical screening.

Key Points

To develop and validate a multimodal model combining clinical variables, radiomics, and deep learning features from computed tomography scans to detect sarcopenia and severe sarcopenia in patients with gastric cancer.
Analyzed 1067 patients with gastric cancer from two medical centers, defining sarcopenia by low grip strength and low skeletal muscle index, with severe sarcopenia additionally requiring low gait speed.
Extracted radiomic features and developed 2D and 2.5D ResNet50 deep learning models from computed tomography images at the third lumbar vertebra level.
Constructed a transformer-based multimodal model integrating clinical data, radiomics, and 2.5D deep learning features, comparing its performance against single-modality models and an XGBoost fusion model across training, validation, and external test cohorts.
Patients with sarcopenia had significantly worse overall survival compared to those without sarcopenia (p < 0.05).
The transformer-based sarcopenia model achieved AUCs of 0.95 (training), 0.87 (validation), and 0.89 (external test) for sarcopenia, alongside AUCs of 0.93, 0.85, and 0.84 for severe sarcopenia.
Single-modality radiomics achieved validation and external test AUCs of 0.85 and 0.80, the 2.5D deep learning model achieved 0.81 and 0.83, and the comparator XGBoost fusion model achieved 0.85 and 0.86, respectively.

Cite This Study

Zhi et al. (2026) studied this question.

synapsesocial.com/papers/6aa9136e9013453be30a14e1 https://doi.org/10.1002/jcsm.70377

View Full Paper

Ask AI

Bookmark

September 15, 2026Journal of Cachexia Sarcopenia and MuscleOpen Access

Integrated Assessment of Sarcopenia in Patients with Gastric Cancer Using Deep Learning and Radiomics

View Full Paper

Ask AI

Bookmark

Authors

HZHuaiqing Zhi JZJingwei Zheng HCHao Chen

Discussion

Member takes

Overview

Multicenter cohort study demonstrates accurate identification of sarcopenia in gastric cancer using multimodal radiomics and deep learning, indicating potential for automated clinical screening.

Key Points

To develop and validate a multimodal model combining clinical variables, radiomics, and deep learning features from computed tomography scans to detect sarcopenia and severe sarcopenia in patients with gastric cancer.
Analyzed 1067 patients with gastric cancer from two medical centers, defining sarcopenia by low grip strength and low skeletal muscle index, with severe sarcopenia additionally requiring low gait speed.
Extracted radiomic features and developed 2D and 2.5D ResNet50 deep learning models from computed tomography images at the third lumbar vertebra level.
Constructed a transformer-based multimodal model integrating clinical data, radiomics, and 2.5D deep learning features, comparing its performance against single-modality models and an XGBoost fusion model across training, validation, and external test cohorts.
Patients with sarcopenia had significantly worse overall survival compared to those without sarcopenia (p < 0.05).
The transformer-based sarcopenia model achieved AUCs of 0.95 (training), 0.87 (validation), and 0.89 (external test) for sarcopenia, alongside AUCs of 0.93, 0.85, and 0.84 for severe sarcopenia.
Single-modality radiomics achieved validation and external test AUCs of 0.85 and 0.80, the 2.5D deep learning model achieved 0.81 and 0.83, and the comparator XGBoost fusion model achieved 0.85 and 0.86, respectively.

Cite This Study

Zhi et al. (2026) studied this question.

synapsesocial.com/papers/6aa9136e9013453be30a14e1 https://doi.org/10.1002/jcsm.70377

View Full Paper

Ask AI

Bookmark