TY - JOUR
T1 - COVLIAS 2.0-cXAI
T2 - Cloud-Based Explainable Deep Learning System for COVID-19 Lesion Localization in Computed Tomography Scans
AU - Suri, Jasjit S.
AU - Agarwal, Sushant
AU - Chabert, Gian Luca
AU - Carriero, Alessandro
AU - Paschè, Alessio
AU - Danna, Pietro S.C.
AU - Saba, Luca
AU - Mehmedović, Armin
AU - Faa, Gavino
AU - Singh, Inder M.
AU - Turk, Monika
AU - Chadha, Paramjit S.
AU - Johri, Amer M.
AU - Khanna, Narendra N.
AU - Mavrogeni, Sophie
AU - Laird, John R.
AU - Pareek, Gyan
AU - Miner, Martin
AU - Sobel, David W.
AU - Balestrieri, Antonella
AU - Sfikakis, Petros P.
AU - Tsoulfas, George
AU - Protogerou, Athanasios D.
AU - Misra, Durga Prasanna
AU - Agarwal, Vikas
AU - Kitas, George D.
AU - Teji, Jagjit S.
AU - Al-Maini, Mustafa
AU - Dhanjil, Surinder K.
AU - Nicolaides, Andrew
AU - Sharma, Aditya
AU - Rathore, Vijay
AU - Fatemi, Mostafa
AU - Alizad, Azra
AU - Krishnan, Pudukode R.
AU - Nagy, Ferenc
AU - Ruzsa, Zoltan
AU - Fouda, Mostafa M.
AU - Naidu, Subbaram
AU - Viskovic, Klaudija
AU - Kalra, Mannudeep K.
N1 - Publisher Copyright:
Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland.
PY - 2022/6
Y1 - 2022/6
N2 - Background: The previous COVID-19 lung diagnosis system lacks both scientific validation and the role of explainable artificial intelligence (AI) for understanding lesion localization. This study presents a cloud-based explainable AI, the “COVLIAS 2.0-cXAI” system using four kinds of class activation maps (CAM) models. Methodology: Our cohort consisted of ~6000 CT slices from two sources (Croatia, 80 COVID-19 patients and Italy, 15 control patients). COVLIAS 2.0-cXAI design consisted of three stages: (i) automated lung segmentation using hybrid deep learning ResNet-UNet model by automatic adjustment of Hounsfield units, hyperparameter optimization, and parallel and distributed training, (ii) classification using three kinds of DenseNet (DN) models (DN-121, DN-169, DN-201), and (iii) validation using four kinds of CAM visualization techniques: gradient-weighted class activation mapping (Grad-CAM), Grad-CAM++, score-weighted CAM (Score-CAM), and FasterScore-CAM. The COVLIAS 2.0-cXAI was validated by three trained senior radiologists for its stability and reliability. The Friedman test was also performed on the scores of the three radiologists. Results: The ResNet-UNet segmentation model resulted in dice similarity of 0.96, Jaccard index of 0.93, a correlation coefficient of 0.99, with a figure-of-merit of 95.99%, while the classifier accuracies for the three DN nets (DN-121, DN-169, and DN-201) were 98%, 98%, and 99% with a loss of ~0.003, ~0.0025, and ~0.002 using 50 epochs, respectively. The mean AUC for all three DN models was 0.99 (p < 0.0001). The COVLIAS 2.0-cXAI showed 80% scans for mean alignment index (MAI) between heatmaps and gold standard, a score of four out of five, establishing the system for clinical settings. Conclusions: The COVLIAS 2.0-cXAI successfully showed a cloud-based explainable AI system for lesion localization in lung CT scans.
AB - Background: The previous COVID-19 lung diagnosis system lacks both scientific validation and the role of explainable artificial intelligence (AI) for understanding lesion localization. This study presents a cloud-based explainable AI, the “COVLIAS 2.0-cXAI” system using four kinds of class activation maps (CAM) models. Methodology: Our cohort consisted of ~6000 CT slices from two sources (Croatia, 80 COVID-19 patients and Italy, 15 control patients). COVLIAS 2.0-cXAI design consisted of three stages: (i) automated lung segmentation using hybrid deep learning ResNet-UNet model by automatic adjustment of Hounsfield units, hyperparameter optimization, and parallel and distributed training, (ii) classification using three kinds of DenseNet (DN) models (DN-121, DN-169, DN-201), and (iii) validation using four kinds of CAM visualization techniques: gradient-weighted class activation mapping (Grad-CAM), Grad-CAM++, score-weighted CAM (Score-CAM), and FasterScore-CAM. The COVLIAS 2.0-cXAI was validated by three trained senior radiologists for its stability and reliability. The Friedman test was also performed on the scores of the three radiologists. Results: The ResNet-UNet segmentation model resulted in dice similarity of 0.96, Jaccard index of 0.93, a correlation coefficient of 0.99, with a figure-of-merit of 95.99%, while the classifier accuracies for the three DN nets (DN-121, DN-169, and DN-201) were 98%, 98%, and 99% with a loss of ~0.003, ~0.0025, and ~0.002 using 50 epochs, respectively. The mean AUC for all three DN models was 0.99 (p < 0.0001). The COVLIAS 2.0-cXAI showed 80% scans for mean alignment index (MAI) between heatmaps and gold standard, a score of four out of five, establishing the system for clinical settings. Conclusions: The COVLIAS 2.0-cXAI successfully showed a cloud-based explainable AI system for lesion localization in lung CT scans.
KW - COVID-19 lesion
KW - FasterScore-CAM
KW - GRAD-CAM
KW - Grad-CAM++
KW - Hounsfield units
KW - Score-CAM
KW - classification
KW - explainable AI
KW - glass ground opacities
KW - hybrid deep learning
KW - lung CT
KW - segmentation
UR - http://www.scopus.com/inward/record.url?scp=85132542446&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85132542446&partnerID=8YFLogxK
U2 - 10.3390/diagnostics12061482
DO - 10.3390/diagnostics12061482
M3 - Article
AN - SCOPUS:85132542446
SN - 2075-4418
VL - 12
JO - Diagnostics
JF - Diagnostics
IS - 6
M1 - 1482
ER -