Paula M Oestmann1,2,3, Clinton J Wang1,4, Lynn J Savic1,2, Charlie A Hamm1,2, Sophie Stark1,2,5, Isabel Schobert1,2, Bernhard Gebauer2, Todd Schlachter1, MingDe Lin1, Jeffrey C Weinreb1, Ramesh Batra6, David Mulligan6, Xuchen Zhang7, James S Duncan1,4, Julius Chapiro8. 1. Department of Radiology and Biomedical Imaging, Yale School of Medicine, 333 Cedar Street, New Haven, CT, 06520, USA. 2. Institute of Radiology, Berlin Institute of Health, Charité - Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt-Universität, 10117, Berlin, Germany. 3. Faculty of Medicine, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany. 4. Department of Biomedical Engineering, Yale School of Engineering and Applied Science, New Haven, CT, 06520, USA. 5. Faculty of Medicine, Albert-Ludwigs-University Freiburg, Freiburg, Germany. 6. Department of Transplantation and Immunology, 333 Cedar Street, New Haven, CT, 06520, USA. 7. Department of Pathology, Yale School of Medicine, 310 Cedar Street, New Haven, CT, 06520, USA. 8. Department of Radiology and Biomedical Imaging, Yale School of Medicine, 333 Cedar Street, New Haven, CT, 06520, USA. julius.chapiro@yale.edu.
Abstract
OBJECTIVES: To train a deep learning model to differentiate between pathologically proven hepatocellular carcinoma (HCC) and non-HCC lesions including lesions with atypical imaging features on MRI. METHODS: This IRB-approved retrospective study included 118 patients with 150 lesions (93 (62%) HCC and 57 (38%) non-HCC) pathologically confirmed through biopsies (n = 72), resections (n = 29), liver transplants (n = 46), and autopsies (n = 3). Forty-seven percent of HCC lesions showed atypical imaging features (not meeting Liver Imaging Reporting and Data System [LI-RADS] criteria for definitive HCC/LR5). A 3D convolutional neural network (CNN) was trained on 140 lesions and tested for its ability to classify the 10 remaining lesions (5 HCC/5 non-HCC). Performance of the model was averaged over 150 runs with random sub-sampling to provide class-balanced test sets. A lesion grading system was developed to demonstrate the similarity between atypical HCC and non-HCC lesions prone to misclassification by the CNN. RESULTS: The CNN demonstrated an overall accuracy of 87.3%. Sensitivities/specificities for HCC and non-HCC lesions were 92.7%/82.0% and 82.0%/92.7%, respectively. The area under the receiver operating curve was 0.912. CNN's performance was correlated with the lesion grading system, becoming less accurate the more atypical imaging features the lesions showed. CONCLUSION: This study provides proof-of-concept for CNN-based classification of both typical- and atypical-appearing HCC lesions on multi-phasic MRI, utilizing pathologically confirmed lesions as "ground truth." KEY POINTS: • A CNN trained on atypical appearing pathologically proven HCC lesions not meeting LI-RADS criteria for definitive HCC (LR5) can correctly differentiate HCC lesions from other liver malignancies, potentially expanding the role of image-based diagnosis in primary liver cancer with atypical features. • The trained CNN demonstrated an overall accuracy of 87.3% and a computational time of < 3 ms which paves the way for clinical application as a decision support instrument.
OBJECTIVES: To train a deep learning model to differentiate between pathologically proven hepatocellular carcinoma (HCC) and non-HCC lesions including lesions with atypical imaging features on MRI. METHODS: This IRB-approved retrospective study included 118 patients with 150 lesions (93 (62%) HCC and 57 (38%) non-HCC) pathologically confirmed through biopsies (n = 72), resections (n = 29), liver transplants (n = 46), and autopsies (n = 3). Forty-seven percent of HCC lesions showed atypical imaging features (not meeting Liver Imaging Reporting and Data System [LI-RADS] criteria for definitive HCC/LR5). A 3D convolutional neural network (CNN) was trained on 140 lesions and tested for its ability to classify the 10 remaining lesions (5 HCC/5 non-HCC). Performance of the model was averaged over 150 runs with random sub-sampling to provide class-balanced test sets. A lesion grading system was developed to demonstrate the similarity between atypical HCC and non-HCC lesions prone to misclassification by the CNN. RESULTS: The CNN demonstrated an overall accuracy of 87.3%. Sensitivities/specificities for HCC and non-HCC lesions were 92.7%/82.0% and 82.0%/92.7%, respectively. The area under the receiver operating curve was 0.912. CNN's performance was correlated with the lesion grading system, becoming less accurate the more atypical imaging features the lesions showed. CONCLUSION: This study provides proof-of-concept for CNN-based classification of both typical- and atypical-appearing HCC lesions on multi-phasic MRI, utilizing pathologically confirmed lesions as "ground truth." KEY POINTS: • A CNN trained on atypical appearing pathologically proven HCC lesions not meeting LI-RADS criteria for definitive HCC (LR5) can correctly differentiate HCC lesions from other liver malignancies, potentially expanding the role of image-based diagnosis in primary liver cancer with atypical features. • The trained CNN demonstrated an overall accuracy of 87.3% and a computational time of < 3 ms which paves the way for clinical application as a decision support instrument.
Entities:
Keywords:
Carcinoma, hepatocellular; Deep learning; Liver neoplasms; Magnetic resonance imaging; Neural networks, computer
Authors: Matthew S Davenport; Shokoufeh Khalatbari; Peter S C Liu; Katherine E Maturen; Ravi K Kaza; Ashish P Wasnik; Mahmoud M Al-Hawary; Daniel I Glazer; Erica B Stein; Jeet Patel; Deepak K Somashekar; Benjamin L Viglianti; Hero K Hussain Journal: Radiology Date: 2014-02-18 Impact factor: 11.105
Authors: Clinton J Wang; Charlie A Hamm; Lynn J Savic; Marc Ferrante; Isabel Schobert; Todd Schlachter; MingDe Lin; Jeffrey C Weinreb; James S Duncan; Julius Chapiro; Brian Letzen Journal: Eur Radiol Date: 2019-05-15 Impact factor: 5.315
Authors: B Pérez Saborido; J C Menéu Díaz; S Jiménez de Los Galanes; C Loinaz Segurola; M Abradelo de Usera; M Donat Garrido; A Moreno Elola-Olaso; R Gómez Sánz; C Jiménez Romero; I Garcia García; E Moreno González Journal: Transplant Proc Date: 2005-11 Impact factor: 1.066
Authors: Margot Fodor; Florian Primavesi; Eva Braunwarth; Benno Cardini; Thomas Resch; Reto Bale; Daniel Putzer; Benjamin Henninger; Rupert Oberhuber; Manuel Maglione; Christian Margreiter; Stefan Schneeberger; Dietmar Öfner; Stefan Stättner Journal: Eur Surg Date: 2018-05-22 Impact factor: 0.953
Authors: Róbert Stollmayer; Bettina Katalin Budai; Aladár Rónaszéki; Zita Zsombor; Ildikó Kalina; Erika Hartmann; Gábor Tóth; Péter Szoldán; Viktor Bérczi; Pál Maurovich-Horvat; Pál Novák Kaposi Journal: Cells Date: 2022-05-05 Impact factor: 6.600
Authors: Lazaro Hiram Betancourt; Jeovanis Gil; Yonghyo Kim; Viktória Doma; Uğur Çakır; Aniel Sanchez; Jimmy Rodriguez Murillo; Magdalena Kuras; Indira Pla Parada; Yutaka Sugihara; Roger Appelqvist; Elisabet Wieslander; Charlotte Welinder; Erika Velasquez; Natália Pinto de Almeida; Nicole Woldmar; Matilda Marko-Varga; Krzysztof Pawłowski; Jonatan Eriksson; Beáta Szeitz; Bo Baldetorp; Christian Ingvar; Håkan Olsson; Lotta Lundgren; Henrik Lindberg; Henriett Oskolas; Boram Lee; Ethan Berge; Marie Sjögren; Carina Eriksson; Dasol Kim; Ho Jeong Kwon; Beatrice Knudsen; Melinda Rezeli; Runyu Hong; Peter Horvatovich; Tasso Miliotis; Toshihide Nishimura; Harubumi Kato; Erik Steinfelder; Madalina Oppermann; Ken Miller; Francesco Florindi; Qimin Zhou; Gilberto B Domont; Luciana Pizzatti; Fábio C S Nogueira; Peter Horvath; Leticia Szadai; József Tímár; Sarolta Kárpáti; A Marcell Szász; Johan Malm; David Fenyö; Henrik Ekedahl; István Balázs Németh; György Marko-Varga Journal: Clin Transl Med Date: 2021-07