Anita C Chandrasekaran1, Zhicheng Fu2,3, Reid Kraniski4, F Perry Wilson5, Shannon Teaw1, Michelle Cheng1, Annie Wang4, Shangping Ren2,6, Imran M Omar7, Monique E Hinchcliff8,9,10. 1. Yale School of Medicine, Section of Rheumatology, Allergy & Immunology, The Anlyan Center, 300 Cedar Street, PO BOX 208031, New Haven, CT, 06520, USA. 2. Department of Computer Science, Illinois Institute of Technology, 10 W 31st St, Chicago, IL, 60616, USA. 3. Motorola Mobility LLC, 222 W Merchandise Mart Plaza #1800, Chicago, IL, 60654, USA. 4. Department of Radiology, Yale School of Medicine, 330 Cedar St, New Haven, CT, 06520, USA. 5. Clinical and Translational Research Accelerator, Department of Medicine, Yale School of Medicine, Temple Medical Center, 60 Temple Street Suite 6C, New Haven, CT, 06510, USA. 6. Department of Computer Science, San Diego State University, 5500 Campanile Drive, San Diego, CA, 92182, USA. 7. Department of Radiology, Northwestern University Feinberg School of Medicine, 676 N St Clair St, Chicago, IL, 60611, USA. 8. Yale School of Medicine, Section of Rheumatology, Allergy & Immunology, The Anlyan Center, 300 Cedar Street, PO BOX 208031, New Haven, CT, 06520, USA. Monique.hinchcliff@yale.edu. 9. Clinical and Translational Research Accelerator, Department of Medicine, Yale School of Medicine, Temple Medical Center, 60 Temple Street Suite 6C, New Haven, CT, 06510, USA. Monique.hinchcliff@yale.edu. 10. Department of Medicine, Division of Rheumatology, Northwestern University Feinberg School of Medicine, 240 E. Huron Street, Suite M-300, Chicago, IL, 60611, USA. Monique.hinchcliff@yale.edu.
Abstract
BACKGROUND: Although treatments have been proposed for calcinosis cutis (CC) in patients with systemic sclerosis (SSc), a standardized and validated method for CC burden quantification is necessary to enable valid clinical trials. We tested the hypothesis that computer vision applied to dual-energy computed tomography (DECT) finger images is a useful approach for precise and accurate CC quantification in SSc patients. METHODS: De-identified 2-dimensional (2D) DECT images from SSc patients with clinically evident lesser finger CC lesions were obtained. An expert musculoskeletal radiologist confirmed accurate manual segmentation (subtraction) of the phalanges for each image as a gold standard, and a U-Net Convolutional Neural Network (CNN) computer vision model for segmentation of healthy phalanges was developed and tested. A validation study was performed in an independent dataset whereby two independent radiologists manually measured the longest length and perpendicular short axis of each lesion and then calculated an estimated area by assuming the lesion was elliptical using the formula long axis/2 × short axis/2 × π, and a computer scientist used a region growing technique to calculate the area of CC lesions. Spearman's correlation coefficient, Lin's concordance correlation coefficient with 95% confidence intervals (CI), and a Bland-Altman plot (Stata V 15.1, College Station, TX) were used to test for equivalence between the radiologists' and the CNN algorithm-generated area estimates. RESULTS: Forty de-identified 2D DECT images from SSc patients with clinically evident finger CC lesions were obtained and divided into training (N = 30 with image rotation × 3 to expand the set to N = 120) and test sets (N = 10). In the training set, five hundred epochs (iterations) were required to train the CNN algorithm to segment phalanges from adjacent CC, and accurate segmentation was evaluated using the ten held-out images. To test model performance, CC lesional area estimates calculated by two independent radiologists and a computer scientist were compared (radiologist 1 vs. radiologist 2 and radiologist 1 vs. computer vision approach) using an independent test dataset comprised of 31 images (8 index finger and 23 other fingers). For the two radiologists', and the radiologist vs. computer vision measurements, Spearman's rho was 0.91 and 0.94, respectively, both p < 0.0001; Lin's concordance correlation coefficient was 0.91 (95% CI 0.85-0.98, p < 0.001) and 0.95 (95% CI 0.91-0.99, p < 0.001); and Bland-Altman plots demonstrated a mean difference between radiologist vs. radiologist, and radiologist vs. computer vision area estimates of - 0.5 mm2 (95% limits of agreement - 10.0-9.0 mm2) and 1.7 mm2 (95% limits of agreement - 6.0-9.5 mm2, respectively. CONCLUSIONS: We demonstrate that CNN quantification has a high degree of correlation with expert radiologist measurement of finger CC area measurements. Future work will include segmentation of 3-dimensional (3D) images for volumetric and density quantification, as well as validation in larger, independent cohorts.
BACKGROUND: Although treatments have been proposed for calcinosis cutis (CC) in patients with systemic sclerosis (SSc), a standardized and validated method for CC burden quantification is necessary to enable valid clinical trials. We tested the hypothesis that computer vision applied to dual-energy computed tomography (DECT) finger images is a useful approach for precise and accurate CC quantification in SSc patients. METHODS: De-identified 2-dimensional (2D) DECT images from SSc patients with clinically evident lesser finger CC lesions were obtained. An expert musculoskeletal radiologist confirmed accurate manual segmentation (subtraction) of the phalanges for each image as a gold standard, and a U-Net Convolutional Neural Network (CNN) computer vision model for segmentation of healthy phalanges was developed and tested. A validation study was performed in an independent dataset whereby two independent radiologists manually measured the longest length and perpendicular short axis of each lesion and then calculated an estimated area by assuming the lesion was elliptical using the formula long axis/2 × short axis/2 × π, and a computer scientist used a region growing technique to calculate the area of CC lesions. Spearman's correlation coefficient, Lin's concordance correlation coefficient with 95% confidence intervals (CI), and a Bland-Altman plot (Stata V 15.1, College Station, TX) were used to test for equivalence between the radiologists' and the CNN algorithm-generated area estimates. RESULTS: Forty de-identified 2D DECT images from SSc patients with clinically evident finger CC lesions were obtained and divided into training (N = 30 with image rotation × 3 to expand the set to N = 120) and test sets (N = 10). In the training set, five hundred epochs (iterations) were required to train the CNN algorithm to segment phalanges from adjacent CC, and accurate segmentation was evaluated using the ten held-out images. To test model performance, CC lesional area estimates calculated by two independent radiologists and a computer scientist were compared (radiologist 1 vs. radiologist 2 and radiologist 1 vs. computer vision approach) using an independent test dataset comprised of 31 images (8 index finger and 23 other fingers). For the two radiologists', and the radiologist vs. computer vision measurements, Spearman's rho was 0.91 and 0.94, respectively, both p < 0.0001; Lin's concordance correlation coefficient was 0.91 (95% CI 0.85-0.98, p < 0.001) and 0.95 (95% CI 0.91-0.99, p < 0.001); and Bland-Altman plots demonstrated a mean difference between radiologist vs. radiologist, and radiologist vs. computer vision area estimates of - 0.5 mm2 (95% limits of agreement - 10.0-9.0 mm2) and 1.7 mm2 (95% limits of agreement - 6.0-9.5 mm2, respectively. CONCLUSIONS: We demonstrate that CNN quantification has a high degree of correlation with expert radiologist measurement of finger CC area measurements. Future work will include segmentation of 3-dimensional (3D) images for volumetric and density quantification, as well as validation in larger, independent cohorts.
Authors: Lorinda Chung; Antonia Valenzuela; David Fiorentino; Kathryn Stevens; Shufeng Li; Jonathan Harris; Charles Hutchinson; Shervin Assassi; Lorenzo Beretta; Santhanam Lakshminarayanan; Tatiana S Rodriguez-Reyna; Christopher P Denton; Rebecca G Taillefer; Ariane L Herrick; Murray Baron Journal: Arthritis Care Res (Hoboken) Date: 2015-03 Impact factor: 4.794
Authors: Brian Garvey; Barış Türkbey; Hong Truong; Marcelino Bernardo; Senthil Periaswamy; Peter L Choyke Journal: Diagn Interv Radiol Date: 2014 May-Jun Impact factor: 2.630
Authors: Andre Esteva; Brett Kuprel; Roberto A Novoa; Justin Ko; Susan M Swetter; Helen M Blau; Sebastian Thrun Journal: Nature Date: 2017-01-25 Impact factor: 49.962
Authors: Thomas George Olsen; B Hunter Jackson; Theresa Ann Feeser; Michael N Kent; John C Moad; Smita Krishnamurthy; Denise D Lunsford; Rajath E Soans Journal: J Pathol Inform Date: 2018-09-27