Mr. Carolus H. J. Kusters Profile

Carolus H. J. Kusters

at Technische Univ. Eindhoven

SPIE Involvement:

Author

Publications (3)

Proceedings Article | 7 April 2023 Poster + Paper

Barrett's lesion detection using a minimal integer-based neural network for embedded systems integration

Tim G. Boers, Carolus H. Kusters, Kiki Fockens, Jelmer Jukema, Martijn Jong, Jeroen de Groof, Jacques Bergman, Fons van der Sommen, Peter H. de With

Proceedings Volume 12465, 1246527 (2023) https://doi.org/10.1117/12.2653890

KEYWORDS: Quantization, Image segmentation, Neural networks, Embedded systems

Read Abstract +

Proceedings Article | 7 April 2023 Poster + Paper

Real-time Barrett's neoplasia characterization in NBI videos using an int8-based quantized neural network

Carolus H. Kusters, Tim G. Boers, Jelmer Jukema, Martijn Jong, Kiki Fockens, Albert de Groof, Jacques Bergman, Fons van der Sommen, Peter H. de With

Proceedings Volume 12465, 124651P (2023) https://doi.org/10.1117/12.2647557

KEYWORDS: Computer aided detection, Endoscopy

Read Abstract +

Proceedings Article | 4 April 2022 Poster + Paper

Colorectal polyp classification using confidence-calibrated convolutional neural networks

Koen Kusters, Thom Scheeve, Nikoo Dehghani, Quirine E. van der Zander, Ramon-Michel Schreuder, Ad A. Masclee, Erik Schoon, Fons van der Sommen, Peter H. de With

Proceedings Volume 12033, 120331P (2022) https://doi.org/10.1117/12.2606801

KEYWORDS: Calibration, Reliability, Error control coding, Computer aided diagnosis and therapy, Convolutional neural networks, Systems modeling, Electrochemical etching, In vivo imaging, Endoscopy, Cancer

Read Abstract +

Computer-Aided Diagnosis (CADx) systems for in-vivo characterization of Colorectal Polyps (CRPs) which are precursor lesions of Colorectal Cancer (CRC), can assist clinicians with diagnosis and better informed decisionmaking during colonoscopy procedures. Current deep learning-based state-of-the-art solutions achieve a high classification performance, but lack measures to increase the reliability of such systems. In this paper, the reliability of a Convolutional Neural Network (CNN) for characterization of CRPs is specifically addressed by confidence calibration. Well-calibrated models produce classification-confidence scores that reflect the actual correctness likelihood of the model, thereby supporting reliable predictions by trustworthy and informative confidence scores. Two recently proposed trainable calibration methods are explored for CRP classification to calibrate the confidence of the proposed CNN. We show that the confidence-calibration error can be decreased by 33.86% (−0.01648 ± 0.01085), 48.33% (−0.04415 ± 0.01731), 50.57% (−0.11423 ± 0.00680), 61.68% (−0.01553 ± 0.00204) and 48.27% (−0.22074 ± 0.08652) for the Expected Calibration Error (ECE), Average Calibration Error (ACE), Maximum Calibration Error (MCE), Over-Confidence Error (OE) and Cumulative Calibration Error (CUMU), respectively. Moreover, the absolute difference between the average entropy and the expected entropy was considerably reduced by 32.00% (−0.04374 ± 0.01238) on average. Furthermore, even a slightly improved classification performance is observed, compared to the uncalibrated equivalent. The obtained results show that the proposed model for CRP classification with confidence calibration produces better calibrated predictions without sacrificing classification performance. This work shows promising points of engagement towards obtaining reliable and well-calibrated CADx systems for in-vivo polyp characterization, to assist clinicians during colonoscopy procedures.

View contact details

UPDATE YOUR PROFILE

Is this your profile? Update it now.

Sign into your SPIE.org account

Don’t have a profile and want one?

Create an account on SPIE.org

Keywords/Phrases

Search In:

Publication Years