Histopathologic Diagnosis Discrepancies Between Preoperative Endoscopic Forceps Biopsies and Specimens Resected by Endoscopic Submucosal Dissection in Superficial Gastric Neoplasms

An artificial intelligence system for comprehensive pathologic outcome prediction in early gastric cancer through endoscopic image analysis (with video)

BACKGROUND

Accurate prediction of pathologic results for early gastric cancer (EGC) based on endoscopic findings is essential in deciding between endoscopic and surgical resection. This study aimed to develop an artificial intelligence (AI) model to assess comprehensive pathologic characteristics of EGC using white-light endoscopic images and videos.

METHODS

To train the model, we retrospectively collected 4,336 images and prospectively included 153 videos from patients with EGC who underwent endoscopic or surgical resection. The performance of the model was tested and compared to that of 16 endoscopists (nine experts and seven novices) using a mutually exclusive set of 260 images and 10 videos. Finally, we conducted external validation using 436 images and 89 videos from another institution.

RESULTS

After training, the model achieved predictive accuracies of 89.7% for undifferentiated histology, 88.0% for submucosal invasion, 87.9% for lymphovascular invasion (LVI), and 92.7% for lymph node metastasis (LNM), using endoscopic videos. The area under the curve values of the model were 0.992 for undifferentiated histology, 0.902 for submucosal invasion, 0.706 for LVI, and 0.680 for LNM in the test. In addition, the model showed significantly higher accuracy than the experts in predicting undifferentiated histology (92.7% vs. 71.6%), submucosal invasion (87.3% vs. 72.6%), and LNM (87.7% vs. 72.3%). The external validation showed accuracies of 75.6% and 71.9% for undifferentiated histology and submucosal invasion, respectively.

CONCLUSIONS

AI may assist endoscopists with high predictive performance for differentiation status and invasion depth of EGC. Further research is needed to improve the detection of LVI and LNM.

The added value of radiomics from dual-energy spectral CT derived iodine-based material decomposition images in predicting histological grade of gastric cancer

BACKGROUND

The histological differentiation grades of gastric cancer (GC) are closely related to treatment choices and prognostic evaluation. Radiomics from dual-energy spectral CT (DESCT) derived iodine-based material decomposition (IMD) images may have the potential to reflect histological grades.

METHODS

A total of 103 patients with pathologically proven GC (low-grade in 40 patients and high-grade in 63 patients) who underwent preoperative DESCT were enrolled in our study. Radiomic features were extracted from conventional polychromatic (CP) images and IMD images, respectively. Three radiomic predictive models (model-CP, model-IMD, and model-CP-IMD) based on solely CP selected features, IMD selected features and CP coupled with IMD selected features were constructed. The clinicopathological data of the enrolled patients were analyzed. Then, we built a combined model (model-Combine) developed with CP-IMD and clinical features. The performance of these models was evaluated and compared.

RESULTS

Model-CP-IMD achieved better AUC results than both model-CP and model-IMD in both cohorts. Model-Combine, which combined CP-IMD radiomic features, pT stage, and pN stage, yielded the highest AUC values of 0.910 and 0.912 in the training and testing cohorts, respectively. Model-CP-IMD and model-Combine outperformed model-CP according to decision curve analysis.

CONCLUSION

DESCT-based radiomics models showed reliable diagnostic performance in predicting GC histologic differentiation grade. The radiomic features extracted from IMD images showed great promise in terms of enhancing diagnostic performance.