A benchmark for comparing precision medicine methods in thyroid cancer diagnosis using tissue microarrays

Area of Science:

• Biomedical Imaging
• Artificial Intelligence in Medicine
• Precision Medicine

Background:

• Precision medicine aims to optimize therapeutic interventions using advanced knowledge and technology.
• Thyroid cancer diagnosis involves complex analysis of histopathological and immunohistochemical tissue images.

Purpose of the Study:

• To explore building AI models for thyroid cancer diagnosis without pixel-level annotation.
• To predict tumor size, extrathyroidal extension, lymph node metastasis, cancer stage, and BRAF mutation.
• To establish a framework for objective evaluation of automatic patient diagnosis algorithms.

Main Methods:

• Development of a novel framework for objective evaluation of automatic diagnosis algorithms under the IEEE International Symposium on Biomedical Imaging 2017 Grand Challenge.
• Creation of data repositories for tissue microarrays and clinical diagnosis classification of thyroid cancer.
• Comparison of three automatic methods for predicting five clinical outcomes using quantitative evaluation.

Main Results:

• A benchmark for objective evaluation and comparison of automatic diagnosis algorithms was established.
• Detailed quantitative evaluation results for three automatic methods were presented.
• Automatic patient diagnosis in thyroid cancer was identified as a challenging and unsolved problem.

Conclusions:

• The established benchmark facilitates future developments in automatic thyroid cancer diagnosis.
• Availability of datasets and evaluation software encourages further research in the field.
• Despite advancements, fully automatic and accurate diagnosis remains an ongoing challenge.