MixPatch: A New Method for Training Histopathology Image Classifiers
CNN-based image processing has been actively applied to histopathological analysis to detect and classify cancerous tumors automatically. However, CNN-based classifiers generally predict a label with overconfidence, which becomes a serious problem in the medical domain. The objective of this study i...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-06-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/12/6/1493 |
| _version_ | 1797488236779536384 |
|---|---|
| author | Youngjin Park Mujin Kim Murtaza Ashraf Young Sin Ko Mun Yong Yi |
| author_facet | Youngjin Park Mujin Kim Murtaza Ashraf Young Sin Ko Mun Yong Yi |
| author_sort | Youngjin Park |
| collection | DOAJ |
| description | CNN-based image processing has been actively applied to histopathological analysis to detect and classify cancerous tumors automatically. However, CNN-based classifiers generally predict a label with overconfidence, which becomes a serious problem in the medical domain. The objective of this study is to propose a new training method, called MixPatch, designed to improve a CNN-based classifier by specifically addressing the prediction uncertainty problem and examine its effectiveness in improving diagnosis performance in the context of histopathological image analysis. MixPatch generates and uses a new sub-training dataset, which consists of mixed-patches and their predefined ground-truth labels, for every single mini-batch. Mixed-patches are generated using a small size of clean patches confirmed by pathologists while their ground-truth labels are defined using a proportion-based soft labeling method. Our results obtained using a large histopathological image dataset shows that the proposed method performs better and alleviates overconfidence more effectively than any other method examined in the study. More specifically, our model showed 97.06% accuracy, an increase of 1.6% to 12.18%, while achieving 0.76% of expected calibration error, a decrease of 0.6% to 6.3%, over the other models. By specifically considering the mixed-region variation characteristics of histopathology images, MixPatch augments the extant mixed image methods for medical image analysis in which prediction uncertainty is a crucial issue. The proposed method provides a new way to systematically alleviate the overconfidence problem of CNN-based classifiers and improve their prediction accuracy, contributing toward more calibrated and reliable histopathology image analysis. |
| first_indexed | 2024-03-09T23:59:12Z |
| format | Article |
| id | doaj.art-4d9a8cadd80b4fbe82573e65e0126983 |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-03-09T23:59:12Z |
| publishDate | 2022-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-4d9a8cadd80b4fbe82573e65e01269832023-11-23T16:19:05ZengMDPI AGDiagnostics2075-44182022-06-01126149310.3390/diagnostics12061493MixPatch: A New Method for Training Histopathology Image ClassifiersYoungjin Park0Mujin Kim1Murtaza Ashraf2Young Sin Ko3Mun Yong Yi4Department of Industrial & Systems Engineering, Korea Advanced Institute of Science and Technology, Daejeon 34141, KoreaDepartment of Industrial & Systems Engineering, Korea Advanced Institute of Science and Technology, Daejeon 34141, KoreaDepartment of Industrial & Systems Engineering, Korea Advanced Institute of Science and Technology, Daejeon 34141, KoreaPathology Center, Seegene Medical Foundation, Seoul 04805, KoreaDepartment of Industrial & Systems Engineering, Korea Advanced Institute of Science and Technology, Daejeon 34141, KoreaCNN-based image processing has been actively applied to histopathological analysis to detect and classify cancerous tumors automatically. However, CNN-based classifiers generally predict a label with overconfidence, which becomes a serious problem in the medical domain. The objective of this study is to propose a new training method, called MixPatch, designed to improve a CNN-based classifier by specifically addressing the prediction uncertainty problem and examine its effectiveness in improving diagnosis performance in the context of histopathological image analysis. MixPatch generates and uses a new sub-training dataset, which consists of mixed-patches and their predefined ground-truth labels, for every single mini-batch. Mixed-patches are generated using a small size of clean patches confirmed by pathologists while their ground-truth labels are defined using a proportion-based soft labeling method. Our results obtained using a large histopathological image dataset shows that the proposed method performs better and alleviates overconfidence more effectively than any other method examined in the study. More specifically, our model showed 97.06% accuracy, an increase of 1.6% to 12.18%, while achieving 0.76% of expected calibration error, a decrease of 0.6% to 6.3%, over the other models. By specifically considering the mixed-region variation characteristics of histopathology images, MixPatch augments the extant mixed image methods for medical image analysis in which prediction uncertainty is a crucial issue. The proposed method provides a new way to systematically alleviate the overconfidence problem of CNN-based classifiers and improve their prediction accuracy, contributing toward more calibrated and reliable histopathology image analysis.https://www.mdpi.com/2075-4418/12/6/1493histopathology image analysisdeep learningprediction uncertaintyconfidence calibration |
| spellingShingle | Youngjin Park Mujin Kim Murtaza Ashraf Young Sin Ko Mun Yong Yi MixPatch: A New Method for Training Histopathology Image Classifiers Diagnostics histopathology image analysis deep learning prediction uncertainty confidence calibration |
| title | MixPatch: A New Method for Training Histopathology Image Classifiers |
| title_full | MixPatch: A New Method for Training Histopathology Image Classifiers |
| title_fullStr | MixPatch: A New Method for Training Histopathology Image Classifiers |
| title_full_unstemmed | MixPatch: A New Method for Training Histopathology Image Classifiers |
| title_short | MixPatch: A New Method for Training Histopathology Image Classifiers |
| title_sort | mixpatch a new method for training histopathology image classifiers |
| topic | histopathology image analysis deep learning prediction uncertainty confidence calibration |
| url | https://www.mdpi.com/2075-4418/12/6/1493 |
| work_keys_str_mv | AT youngjinpark mixpatchanewmethodfortraininghistopathologyimageclassifiers AT mujinkim mixpatchanewmethodfortraininghistopathologyimageclassifiers AT murtazaashraf mixpatchanewmethodfortraininghistopathologyimageclassifiers AT youngsinko mixpatchanewmethodfortraininghistopathologyimageclassifiers AT munyongyi mixpatchanewmethodfortraininghistopathologyimageclassifiers |