Classification of glomerular pathological findings using deep learning and nephrologist–AI collective intelligence approach

Abstract

Background: Automated classification of glomerular pathological findings is potentially beneficial in establishing an efficient and objective diagnosis in renal pathology. While previous studies have verified the artificial intelligence (AI) models for the classification of global sclerosis and glomerular cell proliferation, there are several other glomerular pathological findings required for diagnosis, and the comprehensive models for the classification of these major findings have not yet been reported. Whether the cooperation between these AI models and clinicians improves diagnostic performance also remains unknown. Here, we developed AI models to classify glomerular images for major findings required for pathological diagnosis and investigated whether those models could improve the diagnostic performance of nephrologists. Methods: We used a dataset of 283 kidney biopsy cases comprising 15888 glomerular images that were annotated by a total of 25 nephrologists. AI models to classify seven pathological findings: global sclerosis, segmental sclerosis, endocapillary proliferation, mesangial matrix accumulation, mesangial cell proliferation, crescent, and basement membrane structural changes, were constructed using deep learning by fine-tuning of InceptionV3 convolutional neural network. Subsequently, we compared the agreement to truth labels between majority decision among nephrologists with or without the AI model as a voter. Results: Our model for global sclerosis showed high performance (area under the curve: periodic acid-Schiff, 0.986; periodic acid methenamine silver, 0.983); the models for the other findings also showed performance close to those of nephrologists. By adding the AI model output to majority decision among nephrologists, the sensitivity and specificity were significantly improved in 9 of 14 constructed models compared to those of nephrologists alone. Conclusion: Our study showed a proof-of-concept for the classification of multiple glomerular findings in a comprehensive method of deep learning and suggested its potential effectiveness in improving diagnostic accuracy of clinicians.

Studying human-AI collaboration protocols: the case of the Kasparov’s law in radiological double reading

“…53, The Bundle of Sticks) inspired countless variations of the old saying running "union is strength" (or "in unity is strength"). Indeed in our study, the protocols involving 3 readers and majority voting, and hence a decision best of three, are easily those associated with the more accurate group performance, in accordance with other recent studies [48]. In Fig.…”

“…[ 27 , 38 , 49 ]) by combining their multiple contributions together, regardless of their human or “machinic” nature (e.g. [ 22 , 48 ]).…”

Smartpathk: a platform for teaching glomerulopathies using machine learning

“…In addition to this project, several other methodologies, mainly using AI algorithms, are being validated for use in the routine work of the nephropathologist's laboratory aiming at better accuracy and speed in the assessment of punctual morphological findings of renal biopsies [9,21,24]; however, there has as yet been no use of such methods in routine laboratory work or for educational support, such as the system developed in this study.…”

“…This can be confirmed by the increase in the number of machine learning applications in renal pathology in recent years [7]. Initial studies using AI in nephropathology validated the method's accuracy in identifying the standard histological lesions common in glomerular diseases, prompting researchers to continue using AI concepts and tools to complement the arsenal for nephropathology diagnosis [9,21], however there is no computer program for universalizing analysis or for teaching nephropathology, yet. As ML applications in renal pathology for educational purposes and medical training are in an incipient stage, so to introduce digital pathology into the classroom, we sought to determine whether ML algorithms have acceptable accuracy in relation to conventional microscopy for the development of a tool that facilitates educational training in the teaching of glomerulopathies.…”

“…Because of all this complexity, strategies have been sought in the digital age to replace the repetitive and exhaustive tasks of the histopathological analysis of the nephropathologist in conventional microscopy with machine learning algorithms [9]. In this direction, the study of pathologies in digital form has been adopted as a new analysis strategy using virtual microscopy, with slides of conventional histological preparations with special stains for optical microscopy and immunofluorescence slides scanned entirely in specialized equipment and displayed on the screens of computers [10][11][12].…”