Detecting multiple sub-types of breast cancer in a single patient

摘要: Determining the molecular or genomic sub-type of a cancer particular organ is important for prognosis and treatment planning. While clinical tests determine dominant subtype in patient, it believed that some treatments targeting sub-types ultimately prove ineffective because they ignore existence additional same patient. We present method to detect co-existence two common breast — HER2 basal-like whole slide images H&E-stained tumors Our main contribution formulating this problem precision medicine preparation its training data. The detectors tested be highly accurate classifying patches taken from test cases with well-separated profiles, confirming image classification was effective detecting sub-types. Interestingly, patients whose profiles were not very clearly separated appeared include heterogeneous other indicating secondary within single detection patient may impact cancer.