Deep compressed pneumonia detection for low-power embedded devices

Hongjia Li; Sheng Lin; Ning Liu; Caiwen Ding; Yanzhi Wang

doi:10.1007/978-3-030-33642-4_10

Deep compressed pneumonia detection for low-power embedded devices

Hongjia Li, Sheng Lin, Ning Liu, Caiwen Ding, Yanzhi Wang

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Deep neural networks (DNNs) have been expanded into medical fields and triggered the revolution of some medical applications by extracting complex features and achieving high accuracy and performance, etc. On the contrast, the large-scale network brings high requirements of both memory storage and computation resource, especially for portable medical devices and other embedded systems. In this work, we first train a DNN for pneumonia detection using the dataset provided by RSNA Pneumonia Detection Challenge [4]. To overcome hardware limitation for implementing large-scale networks, we develop a systematic structured weight pruning method with filter sparsity, column sparsity and combined sparsity. Experiments show that we can achieve up to 36x compression ratio compared to the original model with 106 layers, while maintaining no accuracy degradation. We evaluate the proposed methods on an embedded low-power device, Jetson TX2, and achieve low power usage and high energy efficiency.

Original language	English (US)
Title of host publication	Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings
Editors	Luping Zhou, Nicholas Heller, Yiyu Shi, Danny Chen, X. Sharon Hu, Yiming Xiao, Raphael Sznitman, Veronika Cheplygina, Diana Mateus, Emanuele Trucco, Matthieu Chabanas, Hassan Rivaz, Ingerid Reinertsen
Publisher	Springer
Pages	89-97
Number of pages	9
ISBN (Print)	9783030336417
DOIs	https://doi.org/10.1007/978-3-030-33642-4_10
State	Published - 2019
Externally published	Yes
Event	4th International Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis, LABELS 2019, the 1st International Workshop on Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention, HAL-MICCAI 2019, and the 2nd International Workshop on Correction of Brainshift with Intra-Operative Ultrasound, CuRIOUS 2019, held in conjunction with the 22nd International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2019 - Shenzhen, China Duration: Oct 17 2019 → Oct 17 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11851 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	4th International Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis, LABELS 2019, the 1st International Workshop on Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention, HAL-MICCAI 2019, and the 2nd International Workshop on Correction of Brainshift with Intra-Operative Ultrasound, CuRIOUS 2019, held in conjunction with the 22nd International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2019
Country/Territory	China
City	Shenzhen
Period	10/17/19 → 10/17/19

Bibliographical note

Publisher Copyright:
© Springer Nature Switzerland AG 2019.

Keywords

Pneumonia detection
Structured weight pruning
YOLO

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Publisher link

10.1007/978-3-030-33642-4_10

Find It

Find It at U of M Libraries

Cite this

Li, H., Lin, S., Liu, N., Ding, C., & Wang, Y. (2019). Deep compressed pneumonia detection for low-power embedded devices. In L. Zhou, N. Heller, Y. Shi, D. Chen, X. S. Hu, Y. Xiao, R. Sznitman, V. Cheplygina, D. Mateus, E. Trucco, M. Chabanas, H. Rivaz, & I. Reinertsen (Eds.), Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings (pp. 89-97). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11851 LNCS). Springer. https://doi.org/10.1007/978-3-030-33642-4_10

Deep compressed pneumonia detection for low-power embedded devices. / Li, Hongjia; Lin, Sheng; Liu, Ning et al.
Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings. ed. / Luping Zhou; Nicholas Heller; Yiyu Shi; Danny Chen; X. Sharon Hu; Yiming Xiao; Raphael Sznitman; Veronika Cheplygina; Diana Mateus; Emanuele Trucco; Matthieu Chabanas; Hassan Rivaz; Ingerid Reinertsen. Springer, 2019. p. 89-97 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11851 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Li, H, Lin, S, Liu, N, Ding, C & Wang, Y 2019, Deep compressed pneumonia detection for low-power embedded devices. in L Zhou, N Heller, Y Shi, D Chen, XS Hu, Y Xiao, R Sznitman, V Cheplygina, D Mateus, E Trucco, M Chabanas, H Rivaz & I Reinertsen (eds), Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11851 LNCS, Springer, pp. 89-97, 4th International Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis, LABELS 2019, the 1st International Workshop on Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention, HAL-MICCAI 2019, and the 2nd International Workshop on Correction of Brainshift with Intra-Operative Ultrasound, CuRIOUS 2019, held in conjunction with the 22nd International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2019, Shenzhen, China, 10/17/19. https://doi.org/10.1007/978-3-030-33642-4_10

Li H, Lin S, Liu N, Ding C, Wang Y. Deep compressed pneumonia detection for low-power embedded devices. In Zhou L, Heller N, Shi Y, Chen D, Hu XS, Xiao Y, Sznitman R, Cheplygina V, Mateus D, Trucco E, Chabanas M, Rivaz H, Reinertsen I, editors, Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings. Springer. 2019. p. 89-97. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-33642-4_10

Li, Hongjia ; Lin, Sheng ; Liu, Ning et al. / Deep compressed pneumonia detection for low-power embedded devices. Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings. editor / Luping Zhou ; Nicholas Heller ; Yiyu Shi ; Danny Chen ; X. Sharon Hu ; Yiming Xiao ; Raphael Sznitman ; Veronika Cheplygina ; Diana Mateus ; Emanuele Trucco ; Matthieu Chabanas ; Hassan Rivaz ; Ingerid Reinertsen. Springer, 2019. pp. 89-97 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{d7798a772b40466fbf79369add3970d5,

title = "Deep compressed pneumonia detection for low-power embedded devices",

abstract = "Deep neural networks (DNNs) have been expanded into medical fields and triggered the revolution of some medical applications by extracting complex features and achieving high accuracy and performance, etc. On the contrast, the large-scale network brings high requirements of both memory storage and computation resource, especially for portable medical devices and other embedded systems. In this work, we first train a DNN for pneumonia detection using the dataset provided by RSNA Pneumonia Detection Challenge [4]. To overcome hardware limitation for implementing large-scale networks, we develop a systematic structured weight pruning method with filter sparsity, column sparsity and combined sparsity. Experiments show that we can achieve up to 36x compression ratio compared to the original model with 106 layers, while maintaining no accuracy degradation. We evaluate the proposed methods on an embedded low-power device, Jetson TX2, and achieve low power usage and high energy efficiency.",

keywords = "Pneumonia detection, Structured weight pruning, YOLO",

author = "Hongjia Li and Sheng Lin and Ning Liu and Caiwen Ding and Yanzhi Wang",

note = "Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2019.; 4th International Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis, LABELS 2019, the 1st International Workshop on Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention, HAL-MICCAI 2019, and the 2nd International Workshop on Correction of Brainshift with Intra-Operative Ultrasound, CuRIOUS 2019, held in conjunction with the 22nd International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2019 ; Conference date: 17-10-2019 Through 17-10-2019",

year = "2019",

doi = "10.1007/978-3-030-33642-4_10",

language = "English (US)",

isbn = "9783030336417",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "89--97",

editor = "Luping Zhou and Nicholas Heller and Yiyu Shi and Danny Chen and Hu, {X. Sharon} and Yiming Xiao and Raphael Sznitman and Veronika Cheplygina and Diana Mateus and Emanuele Trucco and Matthieu Chabanas and Hassan Rivaz and Ingerid Reinertsen",

booktitle = "Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings",

}

TY - GEN

T1 - Deep compressed pneumonia detection for low-power embedded devices

AU - Li, Hongjia

AU - Lin, Sheng

AU - Liu, Ning

AU - Ding, Caiwen

AU - Wang, Yanzhi

N1 - Publisher Copyright: © Springer Nature Switzerland AG 2019.

PY - 2019

Y1 - 2019

N2 - Deep neural networks (DNNs) have been expanded into medical fields and triggered the revolution of some medical applications by extracting complex features and achieving high accuracy and performance, etc. On the contrast, the large-scale network brings high requirements of both memory storage and computation resource, especially for portable medical devices and other embedded systems. In this work, we first train a DNN for pneumonia detection using the dataset provided by RSNA Pneumonia Detection Challenge [4]. To overcome hardware limitation for implementing large-scale networks, we develop a systematic structured weight pruning method with filter sparsity, column sparsity and combined sparsity. Experiments show that we can achieve up to 36x compression ratio compared to the original model with 106 layers, while maintaining no accuracy degradation. We evaluate the proposed methods on an embedded low-power device, Jetson TX2, and achieve low power usage and high energy efficiency.

AB - Deep neural networks (DNNs) have been expanded into medical fields and triggered the revolution of some medical applications by extracting complex features and achieving high accuracy and performance, etc. On the contrast, the large-scale network brings high requirements of both memory storage and computation resource, especially for portable medical devices and other embedded systems. In this work, we first train a DNN for pneumonia detection using the dataset provided by RSNA Pneumonia Detection Challenge [4]. To overcome hardware limitation for implementing large-scale networks, we develop a systematic structured weight pruning method with filter sparsity, column sparsity and combined sparsity. Experiments show that we can achieve up to 36x compression ratio compared to the original model with 106 layers, while maintaining no accuracy degradation. We evaluate the proposed methods on an embedded low-power device, Jetson TX2, and achieve low power usage and high energy efficiency.

KW - Pneumonia detection

KW - Structured weight pruning

KW - YOLO

UR - http://www.scopus.com/inward/record.url?scp=85076716867&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85076716867&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-33642-4_10

DO - 10.1007/978-3-030-33642-4_10

M3 - Conference contribution

AN - SCOPUS:85076716867

SN - 9783030336417

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 89

EP - 97

BT - Large-Scale Annotation of Biomedical Data and Expert Label Synthesis and Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention - International Workshops, LABELS 2019, HAL-MICCAI 2019, and CuRIOUS 2019, held in Conjunction with MICCAI 2019, Proceedings

A2 - Zhou, Luping

A2 - Heller, Nicholas

A2 - Shi, Yiyu

A2 - Chen, Danny

A2 - Hu, X. Sharon

A2 - Xiao, Yiming

A2 - Sznitman, Raphael

A2 - Cheplygina, Veronika

A2 - Mateus, Diana

A2 - Trucco, Emanuele

A2 - Chabanas, Matthieu

A2 - Rivaz, Hassan

A2 - Reinertsen, Ingerid

PB - Springer

T2 - 4th International Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis, LABELS 2019, the 1st International Workshop on Hardware Aware Learning for Medical Imaging and Computer Assisted Intervention, HAL-MICCAI 2019, and the 2nd International Workshop on Correction of Brainshift with Intra-Operative Ultrasound, CuRIOUS 2019, held in conjunction with the 22nd International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2019

Y2 - 17 October 2019 through 17 October 2019

ER -

Deep compressed pneumonia detection for low-power embedded devices

Abstract

Publication series

Conference

Bibliographical note

Keywords

UN SDGs

Publisher link

Find It

Other files and links

Fingerprint

Cite this