Melanoma is a type of skin cancer that can be deadly if not detected early. Accurate prognosis is crucial for timely treatment, but current methods are limited and often unreliable. In this study, the authors aimed to develop an image analysis-based prognosis score using Google’s Teachable Machine (TM) in melanoma diagnosis.
Image Analysis
The TM is a deep learning model that can learn from large datasets and generate accurate predictions. The authors used whole slide images (WSIs) of melanoma tumors to train the TM, which can analyze these images and identify features related to prognosis. They developed a scoring system based on these features to predict the likelihood of relapse or death within five years.
Results
The authors tested their prognosis score on 175 WSIs and compared it to the clinical data available for these patients. The results showed a strong correlation between the predicted and actual outcomes, demonstrating the accuracy of the TM-based scoring system. They also validated their method using an independent dataset of 100 WSIs and achieved similar results.
Conclusion
The study demonstrated the potential of the TM-based prognosis score in melanoma diagnosis. The authors believe that this approach can improve accuracy and speed up the diagnostic process, leading to better patient outcomes. The use of AI in histopathology images holds great promise for improving cancer diagnosis and treatment, and further research is warranted to explore its applications in other types of cancer as well.
