Artificial Intelligence in the Diagnosis of Invasive Mold Infection: Development of an Automated Histologic Identification System to Distinguish Between Aspergillus and Mucorales

Background: Histopathological identification is usually required since the sensitivity of fungal culture is not sufficient for accurate diagnosis. On the other hand, pathological diagnosis, especially of molds, often is not accurate, even when performed by an experienced pathologist. This is particularly true in the differentiation between mucormycosis and aspergillosis, which have different drugs of choice and medical management. The diseases can easily become severe in a short period of time in accordance with the severity of the underlying disease or predisposing factors. Therefore, correct diagnosis is extremely important and should be entrusted to the pathologist. Aim: To develop an artificial intelligence (AI)-based automated histological diagnostic system for mold infection to support the diagnosis by general pathologists, especially for distinguishing between Aspergillus and Mucorales. Method: We used two indicators for the diagnostic system; namely, the angle of independent hyphae and tortuosity of each hypha. Results and conclusion: We collected 147 and 67 image samples respectively from standard cases of aspergillosis and mucormycosis. All the images were successfully analyzed by automatic recognition of the two indicators. The independent areas divided by the threshold curve generated by two-dimensional plots of the data clearly include the test data obtained from the cases of Aspergillus and Mucorales. The present study demonstrates the usefulness of our newly developed AI-based diagnostic system. Further investigation is required for its practical use.