Computer aided diagnosis systems are used to assist radiologists in their decision making. The sensitivity of these systems is hindered by the complexity of the structures inside the lungs. Several systems and methods have been proposed to detect and classify lung nodules, but all of them have their strengths and weaknesses. One way to overcome the weaknesses is to combine multiple systems. Systems based on handcrafted features capture a limited set of characteristics from the image, while deep learning based classifiers can deal with a wider range of structures. In this work, several ways to combine a handcrafted feature based classifier with four convolutional neural network are explored. The systems were combined merging the probabilities assigned to the detections in several ways. Support-vector machine, multilayer perceptron and random forest classifiers were used to combine the selected classifiers. The LUNA16 Challenge was used to evaluate the performance of the resulting hybrid systems. In all cases, the hybrid systems outperformed the individual systems. Although the average of sensitivities are similar for most of the combinations, the best hybrid system achieves a gain of 35 extra nodules at 4 FP per scan.
Sóñora Mengana, A, Gonidakis, P, Jansen, B, Garcia Naranjo, JC & Vandemeulebroucke, J 2020, Evaluating several ways to combine handcrafted features-based system with a deep learning system using the LUNA16 Challenge framework. in HK Hahn & MA Mazurowski (eds), SPIE Medical Imaging 2020. vol. 11314, MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS, pp. 113143T1-7, SPIE Medical Imaging 2020, 15/02/20. https://doi.org/10.1117/12.2549778
Sóñora Mengana, A., Gonidakis, P., Jansen, B., Garcia Naranjo, J. C., & Vandemeulebroucke, J. (2020). Evaluating several ways to combine handcrafted features-based system with a deep learning system using the LUNA16 Challenge framework. In H. K. Hahn, & M. A. Mazurowski (Eds.), SPIE Medical Imaging 2020 (Vol. 11314, pp. 113143T1-7). (MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS). https://doi.org/10.1117/12.2549778
@inproceedings{9387a90606a74f70acf4403a308ed87b,
title = "Evaluating several ways to combine handcrafted features-based system with a deep learning system using the LUNA16 Challenge framework",
abstract = "Computer aided diagnosis systems are used to assist radiologists in their decision making. The sensitivity of these systems is hindered by the complexity of the structures inside the lungs. Several systems and methods have been proposed to detect and classify lung nodules, but all of them have their strengths and weaknesses. One way to overcome the weaknesses is to combine multiple systems. Systems based on handcrafted features capture a limited set of characteristics from the image, while deep learning based classifiers can deal with a wider range of structures. In this work, several ways to combine a handcrafted feature based classifier with four convolutional neural network are explored. The systems were combined merging the probabilities assigned to the detections in several ways. Support-vector machine, multilayer perceptron and random forest classifiers were used to combine the selected classifiers. The LUNA16 Challenge was used to evaluate the performance of the resulting hybrid systems. In all cases, the hybrid systems outperformed the individual systems. Although the average of sensitivities are similar for most of the combinations, the best hybrid system achieves a gain of 35 extra nodules at 4 FP per scan.",
author = "{S{\'o}{\~n}ora Mengana}, Alexander and Panagiotis Gonidakis and Bart Jansen and {Garcia Naranjo}, {Juan Carlos} and Jef Vandemeulebroucke",
year = "2020",
month = mar,
doi = "10.1117/12.2549778",
language = "English",
volume = "11314",
series = "MEDICAL IMAGING 2020: COMPUTER-AIDED DIAGNOSIS",
pages = "113143T1--7",
editor = "Hahn, {Horst K.} and Mazurowski, {Maciej A.}",
booktitle = "SPIE Medical Imaging 2020",
note = "SPIE Medical Imaging 2020 ; Conference date: 15-02-2020 Through 20-02-2020",
}