Professor Hsiang-Chen Wang of the National Chung Cheng University (CCU) Department of Mechanical Engineering develops an effective esophageal cancer stage detection method with an accuracy of up to 85%

This is an image  This is an image

Due to the difficulty of detecting esophageal cancer in its early stages, many patients have already entered the advanced stages of this cancer at their time of diagnosis. In light of this problem, Prof. Hsiang-Chen Wang of the National Chung Cheng University (CCU) Department of Mechanical Engineering proposed an analytical method to identify the characteristic spectra of the esophagus through endoscopic imaging. The method not only assists doctors in identifying pathological changes in esophageal cancer, but also allows them to determine the stage of a patient’s esophageal cancer. With an accuracy of up to 85%, the method has been patented in Taiwan and the US, and was also published in the international journal Scientific Reports. Prof. Wang hopes to integrate the method with artificial intelligence, so as to assist doctors in making real-time diagnoses and reduce their stress.

Similar to other “invisible killer” cancers, signs of physical discomfort are not present during the early stages of esophageal cancer. Moreover, since the effectiveness of early esophageal cancer detection is not as high as that for other diseases, most patients have already entered the advanced stages of the cancer at the time of diagnosis, thus delaying opportunities for effective treatment. According to Prof. Wang, in conventional endoscopy, superficial (early stage) esophageal cancer lesions usually present with unclear boundaries and are easily ignored as a result. Moreover, it is sometimes impossible to differentiate between a lesion and a normal esophageal tissue.

According to Prof. Wang, “At present, endoscopic imaging has been used in relevant studies to detect lesions. On the other hand, our technique uses spectral imaging.” In conventional hospitals, endoscopy is used to diagnose esophageal cancer, and doctors mostly rely on visual inspection and empirical diagnosis, which makes it difficult to accurately determine early pathological changes in esophageal cancer. To overcome this problem, Prof. Wang joined hands with a research team led by Dr. I-Chen Wu from the Kaohsiung Medical University Chung-Ho Memorial Hospital. The team from CCU developed an analytical method to identify the characteristic spectra of the esophagus based on 220 endoscopic images of esophageal cancer patients provided by the hospital. The endoscopic images can be used to accurately determine the condition of patient’s esophagus, which saves the need for a biopsy. Furthermore, the method allows doctors to determine if a patient’s cancer is mild, moderate, or severe, and has an accuracy of up to 85%.

Endoscopy techniques include standard white light, iodine staining, and narrow-band imaging. The technique developed by the CCU team combines all three imaging techniques with other optical detection methods such as hyperspectral imaging and principal component analysis, thereby generating spectral data. Prof. Wang says, “The waveforms displayed in the spectral data of different cancer stages have their own distinct characteristics.” Therefore, these spectral data can be used to identify more cancer stages or types.

Statistics reveal that the mortality rate of esophageal cancer is ranked ninth among the ten leading cancers in Taiwan. However, its relatively low incidence rate affects the sample sizes of studies. According to Prof. Wang, the hyperspectral imaging technique also allows for a single pathological image to yield 400 additional images, thus enlarging the sample size and overcoming the problem of insufficient data. Furthermore, the instruments developed for this technique boast high detection rates and user-friendly human machine interfaces, and cost less than typical hyperspectral imaging instruments.

At present, Prof. Wang is still working to improve the accuracy of the analytical technique and plans to integrate the technique with artificial intelligence. If he succeeds, the accuracy of the technique is expected to reach 99.9%. For doctors, this is particularly useful for performing real-time data comparisons during screenings and administering proper treatments to their patients. In addition, the technique can be used to train medical students on the detection and diagnosis of esophageal cancer. The research achievements of Prof. Wang’s team were not only patented in the US and Taiwan, but were also published in Scientific Reports, an international journal of Nature Publishing Group in the UK.

According to CCU, Prof. Wang has studied hyperspectral imaging techniques for 11 years and has applied such techniques to the detection of oral cancer, skin cancer, and macular degeneration. Furthermore, Prof. Wang has garnered success in the field of two-dimensional materials, particularly in membrane-based detection. Recently, he has applied and developed relevant techniques for companies in science parks, thereby increasing Taiwan’s competitiveness in automated optical detection techniques.