Improving utility of brain tumor confocal laser endomicroscopy

Objective value assessment and diagnostic frame detection with convolutional neural networks

Mohammadhassan Izadyyazdanabadi, Evgenii Belykh, Nikolay Martirosyan, Jennifer Eschbacher, Peter Nakaji, Yezhou Yang, Mark C. Preul

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Confocal laser endomicroscopy (CLE), although capable of obtaining images at cellular resolution during surgery of brain tumors in real time, creates as many non-diagnostic as diagnostic images. Non-useful images are often distorted due to relative motion between probe and brain or blood artifacts. Many images, however, simply lack diagnostic features immediately informative to the physician. Examining all the hundreds or thousands of images from a single case to discriminate diagnostic images from nondiagnostic ones can be tedious. Providing a real time diagnostic value assessment of images (fast enough to be used during the surgical acquisition process and accurate enough for the pathologist to rely on) to automatically detect diagnostic frames would streamline the analysis of images and filter useful images for the pathologist/surgeon. We sought to automatically classify images as diagnostic or non-diagnostic. AlexNet, a deep-learning architecture, was used in a 4-fold cross validation manner. Our dataset includes 16,795 images (8572 nondiagnostic and 8223 diagnostic) from 74 CLE-aided brain tumor surgery patients. The ground truth for all the images is provided by the pathologist. Average model accuracy on test data was 91% overall (90.79 % accuracy, 90.94 % sensitivity and 90.87 % specificity). To evaluate the model reliability we also performed receiver operating characteristic (ROC) analysis yielding 0.958 average for area under ROC curve (AUC). These results demonstrate that a deeply trained AlexNet network can achieve a model that reliably and quickly recognizes diagnostic CLE images.

Original languageEnglish (US)
Title of host publicationMedical Imaging 2017
Subtitle of host publicationComputer-Aided Diagnosis
PublisherSPIE
Volume10134
ISBN (Electronic)9781510607132
DOIs
StatePublished - Jan 1 2017
EventMedical Imaging 2017: Computer-Aided Diagnosis - Orlando, United States
Duration: Feb 13 2017Feb 16 2017

Other

OtherMedical Imaging 2017: Computer-Aided Diagnosis
CountryUnited States
CityOrlando
Period2/13/172/16/17

Fingerprint

Brain Neoplasms
brain
Tumors
Brain
Lasers
tumors
Neural networks
ROC Curve
Surgery
lasers
Artifacts
Blood
Learning
Physicians
Sensitivity and Specificity
surgery
Pathologists
receivers
image filters
surgeons

Keywords

  • Brain tumor surgery
  • Computer aided diagnosis
  • Confocal laser endomicroscopy
  • Convolutional neural networks
  • Image quality assessment
  • Precision surgery

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Atomic and Molecular Physics, and Optics
  • Biomaterials
  • Radiology Nuclear Medicine and imaging

Cite this

Izadyyazdanabadi, M., Belykh, E., Martirosyan, N., Eschbacher, J., Nakaji, P., Yang, Y., & Preul, M. C. (2017). Improving utility of brain tumor confocal laser endomicroscopy: Objective value assessment and diagnostic frame detection with convolutional neural networks. In Medical Imaging 2017: Computer-Aided Diagnosis (Vol. 10134). [101342J] SPIE. https://doi.org/10.1117/12.2254902

Improving utility of brain tumor confocal laser endomicroscopy : Objective value assessment and diagnostic frame detection with convolutional neural networks. / Izadyyazdanabadi, Mohammadhassan; Belykh, Evgenii; Martirosyan, Nikolay; Eschbacher, Jennifer; Nakaji, Peter; Yang, Yezhou; Preul, Mark C.

Medical Imaging 2017: Computer-Aided Diagnosis. Vol. 10134 SPIE, 2017. 101342J.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Izadyyazdanabadi, M, Belykh, E, Martirosyan, N, Eschbacher, J, Nakaji, P, Yang, Y & Preul, MC 2017, Improving utility of brain tumor confocal laser endomicroscopy: Objective value assessment and diagnostic frame detection with convolutional neural networks. in Medical Imaging 2017: Computer-Aided Diagnosis. vol. 10134, 101342J, SPIE, Medical Imaging 2017: Computer-Aided Diagnosis, Orlando, United States, 2/13/17. https://doi.org/10.1117/12.2254902
Izadyyazdanabadi M, Belykh E, Martirosyan N, Eschbacher J, Nakaji P, Yang Y et al. Improving utility of brain tumor confocal laser endomicroscopy: Objective value assessment and diagnostic frame detection with convolutional neural networks. In Medical Imaging 2017: Computer-Aided Diagnosis. Vol. 10134. SPIE. 2017. 101342J https://doi.org/10.1117/12.2254902
Izadyyazdanabadi, Mohammadhassan ; Belykh, Evgenii ; Martirosyan, Nikolay ; Eschbacher, Jennifer ; Nakaji, Peter ; Yang, Yezhou ; Preul, Mark C. / Improving utility of brain tumor confocal laser endomicroscopy : Objective value assessment and diagnostic frame detection with convolutional neural networks. Medical Imaging 2017: Computer-Aided Diagnosis. Vol. 10134 SPIE, 2017.
@inproceedings{df283a2ac10848d793c00c8b0730a521,
title = "Improving utility of brain tumor confocal laser endomicroscopy: Objective value assessment and diagnostic frame detection with convolutional neural networks",
abstract = "Confocal laser endomicroscopy (CLE), although capable of obtaining images at cellular resolution during surgery of brain tumors in real time, creates as many non-diagnostic as diagnostic images. Non-useful images are often distorted due to relative motion between probe and brain or blood artifacts. Many images, however, simply lack diagnostic features immediately informative to the physician. Examining all the hundreds or thousands of images from a single case to discriminate diagnostic images from nondiagnostic ones can be tedious. Providing a real time diagnostic value assessment of images (fast enough to be used during the surgical acquisition process and accurate enough for the pathologist to rely on) to automatically detect diagnostic frames would streamline the analysis of images and filter useful images for the pathologist/surgeon. We sought to automatically classify images as diagnostic or non-diagnostic. AlexNet, a deep-learning architecture, was used in a 4-fold cross validation manner. Our dataset includes 16,795 images (8572 nondiagnostic and 8223 diagnostic) from 74 CLE-aided brain tumor surgery patients. The ground truth for all the images is provided by the pathologist. Average model accuracy on test data was 91{\%} overall (90.79 {\%} accuracy, 90.94 {\%} sensitivity and 90.87 {\%} specificity). To evaluate the model reliability we also performed receiver operating characteristic (ROC) analysis yielding 0.958 average for area under ROC curve (AUC). These results demonstrate that a deeply trained AlexNet network can achieve a model that reliably and quickly recognizes diagnostic CLE images.",
keywords = "Brain tumor surgery, Computer aided diagnosis, Confocal laser endomicroscopy, Convolutional neural networks, Image quality assessment, Precision surgery",
author = "Mohammadhassan Izadyyazdanabadi and Evgenii Belykh and Nikolay Martirosyan and Jennifer Eschbacher and Peter Nakaji and Yezhou Yang and Preul, {Mark C.}",
year = "2017",
month = "1",
day = "1",
doi = "10.1117/12.2254902",
language = "English (US)",
volume = "10134",
booktitle = "Medical Imaging 2017",
publisher = "SPIE",
address = "United States",

}

TY - GEN

T1 - Improving utility of brain tumor confocal laser endomicroscopy

T2 - Objective value assessment and diagnostic frame detection with convolutional neural networks

AU - Izadyyazdanabadi, Mohammadhassan

AU - Belykh, Evgenii

AU - Martirosyan, Nikolay

AU - Eschbacher, Jennifer

AU - Nakaji, Peter

AU - Yang, Yezhou

AU - Preul, Mark C.

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Confocal laser endomicroscopy (CLE), although capable of obtaining images at cellular resolution during surgery of brain tumors in real time, creates as many non-diagnostic as diagnostic images. Non-useful images are often distorted due to relative motion between probe and brain or blood artifacts. Many images, however, simply lack diagnostic features immediately informative to the physician. Examining all the hundreds or thousands of images from a single case to discriminate diagnostic images from nondiagnostic ones can be tedious. Providing a real time diagnostic value assessment of images (fast enough to be used during the surgical acquisition process and accurate enough for the pathologist to rely on) to automatically detect diagnostic frames would streamline the analysis of images and filter useful images for the pathologist/surgeon. We sought to automatically classify images as diagnostic or non-diagnostic. AlexNet, a deep-learning architecture, was used in a 4-fold cross validation manner. Our dataset includes 16,795 images (8572 nondiagnostic and 8223 diagnostic) from 74 CLE-aided brain tumor surgery patients. The ground truth for all the images is provided by the pathologist. Average model accuracy on test data was 91% overall (90.79 % accuracy, 90.94 % sensitivity and 90.87 % specificity). To evaluate the model reliability we also performed receiver operating characteristic (ROC) analysis yielding 0.958 average for area under ROC curve (AUC). These results demonstrate that a deeply trained AlexNet network can achieve a model that reliably and quickly recognizes diagnostic CLE images.

AB - Confocal laser endomicroscopy (CLE), although capable of obtaining images at cellular resolution during surgery of brain tumors in real time, creates as many non-diagnostic as diagnostic images. Non-useful images are often distorted due to relative motion between probe and brain or blood artifacts. Many images, however, simply lack diagnostic features immediately informative to the physician. Examining all the hundreds or thousands of images from a single case to discriminate diagnostic images from nondiagnostic ones can be tedious. Providing a real time diagnostic value assessment of images (fast enough to be used during the surgical acquisition process and accurate enough for the pathologist to rely on) to automatically detect diagnostic frames would streamline the analysis of images and filter useful images for the pathologist/surgeon. We sought to automatically classify images as diagnostic or non-diagnostic. AlexNet, a deep-learning architecture, was used in a 4-fold cross validation manner. Our dataset includes 16,795 images (8572 nondiagnostic and 8223 diagnostic) from 74 CLE-aided brain tumor surgery patients. The ground truth for all the images is provided by the pathologist. Average model accuracy on test data was 91% overall (90.79 % accuracy, 90.94 % sensitivity and 90.87 % specificity). To evaluate the model reliability we also performed receiver operating characteristic (ROC) analysis yielding 0.958 average for area under ROC curve (AUC). These results demonstrate that a deeply trained AlexNet network can achieve a model that reliably and quickly recognizes diagnostic CLE images.

KW - Brain tumor surgery

KW - Computer aided diagnosis

KW - Confocal laser endomicroscopy

KW - Convolutional neural networks

KW - Image quality assessment

KW - Precision surgery

UR - http://www.scopus.com/inward/record.url?scp=85020292335&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85020292335&partnerID=8YFLogxK

U2 - 10.1117/12.2254902

DO - 10.1117/12.2254902

M3 - Conference contribution

VL - 10134

BT - Medical Imaging 2017

PB - SPIE

ER -