Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks
Chronic kidney damage is routinely assessed semiquantitatively by scoring the amount of fibrosis and tubular atrophy in a renal biopsy sample. Although image digitization and morphometric techniques can better quantify the extent of histologic damage, we need more widely applicable ways to stratify...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2018-03-01
|
| Series: | Kidney International Reports |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2468024917304370 |
| _version_ | 1828770464099467264 |
|---|---|
| author | Vijaya B. Kolachalama Priyamvada Singh Christopher Q. Lin Dan Mun Mostafa E. Belghasem Joel M. Henderson Jean M. Francis David J. Salant Vipul C. Chitalia |
| author_facet | Vijaya B. Kolachalama Priyamvada Singh Christopher Q. Lin Dan Mun Mostafa E. Belghasem Joel M. Henderson Jean M. Francis David J. Salant Vipul C. Chitalia |
| author_sort | Vijaya B. Kolachalama |
| collection | DOAJ |
| description | Chronic kidney damage is routinely assessed semiquantitatively by scoring the amount of fibrosis and tubular atrophy in a renal biopsy sample. Although image digitization and morphometric techniques can better quantify the extent of histologic damage, we need more widely applicable ways to stratify kidney disease severity.
Methods: We leveraged a deep learning architecture to better associate patient-specific histologic images with clinical phenotypes (training classes) including chronic kidney disease (CKD) stage, serum creatinine, and nephrotic-range proteinuria at the time of biopsy, and 1-, 3-, and 5-year renal survival. Trichrome-stained images processed from renal biopsy samples were collected on 171 patients treated at the Boston Medical Center from 2009 to 2012. Six convolutional neural network (CNN) models were trained using these images as inputs and the training classes as outputs, respectively. For comparison, we also trained separate classifiers using the pathologist-estimated fibrosis score (PEFS) as input and the training classes as outputs, respectively.
Results: CNN models outperformed PEFS across the classification tasks. Specifically, the CNN model predicted the CKD stage more accurately than the PEFS model (κ = 0.519 vs. 0.051). For creatinine models, the area under curve (AUC) was 0.912 (CNN) versus 0.840 (PEFS). For proteinuria models, AUC was 0.867 (CNN) versus 0.702 (PEFS). AUC values for the CNN models for 1-, 3-, and 5-year renal survival were 0.878, 0.875, and 0.904, respectively, whereas the AUC values for PEFS model were 0.811, 0.800, and 0.786, respectively.
Conclusion: The study demonstrates a proof of principle that deep learning can be applied to routine renal biopsy images. |
| first_indexed | 2024-12-11T14:07:35Z |
| format | Article |
| id | doaj.art-c19f2e5332cd44159bb60d362b25d817 |
| institution | Directory Open Access Journal |
| issn | 2468-0249 |
| language | English |
| last_indexed | 2024-12-11T14:07:35Z |
| publishDate | 2018-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Kidney International Reports |
| spelling | doaj.art-c19f2e5332cd44159bb60d362b25d8172022-12-22T01:03:36ZengElsevierKidney International Reports2468-02492018-03-013246447510.1016/j.ekir.2017.11.002Association of Pathological Fibrosis With Renal Survival Using Deep Neural NetworksVijaya B. Kolachalama0Priyamvada Singh1Christopher Q. Lin2Dan Mun3Mostafa E. Belghasem4Joel M. Henderson5Jean M. Francis6David J. Salant7Vipul C. Chitalia8Section of Computational Biomedicine, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USARenal Section, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USACollege of Engineering, Boston University, Boston, Massachusetts, USACollege of Health & Rehabilitation Sciences: Sargent College, Boston University, Boston, Massachusetts, USADepartment of Pathology and Laboratory Medicine, Boston University School of Medicine, Boston, Massachusetts, USADepartment of Pathology and Laboratory Medicine, Boston University School of Medicine, Boston, Massachusetts, USARenal Section, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USARenal Section, Department of Medicine, Boston University School of Medicine, Boston, Massachusetts, USAWhitaker Cardiovascular Institute, Boston University School of Medicine, Boston, Massachusetts, USAChronic kidney damage is routinely assessed semiquantitatively by scoring the amount of fibrosis and tubular atrophy in a renal biopsy sample. Although image digitization and morphometric techniques can better quantify the extent of histologic damage, we need more widely applicable ways to stratify kidney disease severity. Methods: We leveraged a deep learning architecture to better associate patient-specific histologic images with clinical phenotypes (training classes) including chronic kidney disease (CKD) stage, serum creatinine, and nephrotic-range proteinuria at the time of biopsy, and 1-, 3-, and 5-year renal survival. Trichrome-stained images processed from renal biopsy samples were collected on 171 patients treated at the Boston Medical Center from 2009 to 2012. Six convolutional neural network (CNN) models were trained using these images as inputs and the training classes as outputs, respectively. For comparison, we also trained separate classifiers using the pathologist-estimated fibrosis score (PEFS) as input and the training classes as outputs, respectively. Results: CNN models outperformed PEFS across the classification tasks. Specifically, the CNN model predicted the CKD stage more accurately than the PEFS model (κ = 0.519 vs. 0.051). For creatinine models, the area under curve (AUC) was 0.912 (CNN) versus 0.840 (PEFS). For proteinuria models, AUC was 0.867 (CNN) versus 0.702 (PEFS). AUC values for the CNN models for 1-, 3-, and 5-year renal survival were 0.878, 0.875, and 0.904, respectively, whereas the AUC values for PEFS model were 0.811, 0.800, and 0.786, respectively. Conclusion: The study demonstrates a proof of principle that deep learning can be applied to routine renal biopsy images.http://www.sciencedirect.com/science/article/pii/S2468024917304370histologymachine learningrenal fibrosisrenal survival |
| spellingShingle | Vijaya B. Kolachalama Priyamvada Singh Christopher Q. Lin Dan Mun Mostafa E. Belghasem Joel M. Henderson Jean M. Francis David J. Salant Vipul C. Chitalia Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks Kidney International Reports histology machine learning renal fibrosis renal survival |
| title | Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks |
| title_full | Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks |
| title_fullStr | Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks |
| title_full_unstemmed | Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks |
| title_short | Association of Pathological Fibrosis With Renal Survival Using Deep Neural Networks |
| title_sort | association of pathological fibrosis with renal survival using deep neural networks |
| topic | histology machine learning renal fibrosis renal survival |
| url | http://www.sciencedirect.com/science/article/pii/S2468024917304370 |
| work_keys_str_mv | AT vijayabkolachalama associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT priyamvadasingh associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT christopherqlin associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT danmun associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT mostafaebelghasem associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT joelmhenderson associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT jeanmfrancis associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT davidjsalant associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks AT vipulcchitalia associationofpathologicalfibrosiswithrenalsurvivalusingdeepneuralnetworks |