AI Model Enhances Early Gastric Cancer Detection –

4 minutes, 11 seconds Read

Key Takeaways

  • Artificial intelligence (AI) is showing great promise in assisting doctors with finding early stomach cancer.
  • Newer systems go beyond just identifying EGC and can give information about the cancerous area’s size and shape.
  • Future AI advancements could also help determine treatment plans.

Artificial intelligence (AI) is making waves in healthcare, particularly in aiding doctors with early detection of stomach cancer. Researchers are developing AI systems that analyze endoscopic images to identify signs of early gastric cancer (EGC)  with impressive accuracy.

Studies like those  by Hirasawa and Sakai achieved sensitivity rates exceeding 90% in detecting EGC using deep learning techniques. This translates to a high probability of correctly identifying EGC during an endoscopy procedure.

AI Shows Promise In Early Detection

The focus has recently shifted toward AI systems that not only detect EGC but also provide more detailed information about cancerous lesions. Researchers Ishioka, Yoon, and Ueyama all achieved high accuracy in EGC detection, with some models exceeding 98%. Ueyama’s system even achieved near-perfect specificity, meaning it rarely identified healthy tissue as cancerous.

Other research  like Oura’s and Wu’s explored AI systems that outperform human endoscopists in identifying EGC. This technology has the potential to improve diagnostic accuracy and reduce missed diagnoses.

While current AI systems excel at detection, they often rely on particular indicators – rectangular bounding boxes – to highlight lesions, which don’t provide precise information about the cancerous area. Researchers are addressing this by developing AI models for lesion segmentation, such as the IMR-CNN model. This model goes beyond detection by creating a more precise map of the cancerous region within the stomach.

Overall, these studies highlight the immense potential of AI in assisting doctors with early and accurate EGC detection. With continued development, AI could become a valuable tool for improving patient outcomes in stomach cancer diagnosis.

AI Role In “Pathology Identification”

Utilizing artificial intelligence for the detection and management of upper gastrointestinal, small bowel, and lower gastrointestinal cancers, defines and underscores the anticipated value of AI in healthcare. Gastrointestinal endoscopy comprises seven distinct stages: pre-procedure, procedure completion, pathology identification, pathology management, handling of complications, patient experience, and post-procedure follow-up. Presently, AI finds its most common application in the stage of “pathology identification.”

In the upper gastrointestinal tract, AI technology aids in detecting gastric neoplasia and predicting submucosal invasion during endoscopic procedures. The reported miss rate for gastric tumors is 10%, primarily attributed to inadequate training exposure due to the low incidence of gastric cancer in Caucasians and incomplete examinations caused by subtle mucosal lesions. AI-supported endoscopy holds promise in addressing these challenges.

An offline study  employing video and still images achieved a sensitivity of 88% and a specificity of 89%. Consequently, AI facilitates risk assessment and treatment planning for gastric neoplastic lesions by estimating lesion type and invasion depth.

While skilled endoscopists can perform endoscopic treatments like endoscopic submucosal dissection (ESD) to identify and cure early lesions, approximately 20% of lesions involve judgment factors such as color changes, redness, nodularity, interruption, convergence of gastric folds, and friability, which may lead to uncertainty regarding complete cure. Herein lies the potential for AI technology to serve as a valuable adjunct, offering a reliable alternative to aid in selecting comprehensive treatment strategies for gastric lesions.

Future Perspectives

The emergence of deep learning technology in the medical field is poised to profoundly impact human quality of life, extending beyond patients to encompass both endoscopic diagnosis outside the operating room and surgical procedures within it. Advances in diagnosing gastrointestinal lesions from esophagus to large intestine are rapidly progressing.

Global development of AI-assisted endoscopic diagnostic technology for early gastric cancer detection shows promising results. Progress is also seen in technology to determine lesion resection margin and depth of invasion, crucial for practical clinical use.

The benefits accruable to endoscopists and surgeons are manifold. Firstly, both novice and expert practitioners can reduce misdiagnosis rates. While diagnosing EGC demands specialized expertise, various variables such as insufficient clinical data, inadequate medical facilities, or occasional lapses in concentration can compromise accuracy. AI technology, providing objective auxiliary diagnoses irrespective of environmental factors, can potentially mitigate false diagnosis rates.

Secondly, by leveraging comprehensive datasets, it can support doctors in challenging scenarios, including those lacking in medical education and experience, underserved regions, and diverse patient populations. Thirdly, it has the potential to alleviate physician burnout by reducing workload.

Although technologies like magnifying endoscopy with narrow-band imaging (ME-NBI) offer enhanced diagnostic accuracy, they demand advanced skills and contribute to fatigue. AI-assisted diagnostic endoscopy has the potential to alleviate physician fatigue and streamline the diagnostic process, thereby reducing endoscopy and procedure times.

For patients, the foremost advantage is the possibility of receiving more tailored treatments. Utilizing surgical resection margins appropriate for various lesion types can minimize side effects and bleeding. Secondly, in regions with limited medical resources or facing issues like religious or gender discrimination, patients stand to receive more objective and improved treatment. Thirdly, it promotes treatment process transparency, enabling patients to comprehend treatment outcomes visually and fostering cooperation to maximize therapeutic efficacy.

Was this Article helpful?



This post was originally published on this site

Similar Posts