Deep learning for FTIR histology

leveraging spatial and spectral features with convolutional neural networks

Sebastian Berisha, Mahsa Lotfollahi, Jahandar Jahanipour, Ilker Gurcan, Michael Walsh, Rohit Bhargava, Hien Van Nguyen, David Mayerich

Research output: Contribution to journalArticle

Abstract

Current methods for cancer detection rely on tissue biopsy, chemical labeling/staining, and examination of the tissue by a pathologist. Though these methods continue to remain the gold standard, they are non-quantitative and susceptible to human error. Fourier transform infrared (FTIR) spectroscopic imaging has shown potential as a quantitative alternative to traditional histology. However, identification of histological components requires reliable classification based on molecular spectra, which are susceptible to artifacts introduced by noise and scattering. Several tissue types, particularly in heterogeneous tissue regions, tend to confound traditional classification methods. Convolutional neural networks (CNNs) are the current state-of-the-art in image classification, providing the ability to learn spatial characteristics of images. In this paper, we demonstrate that CNNs with architectures designed to process both spectral and spatial information can significantly improve classifier performance over per-pixel spectral classification. We report classification results after applying CNNs to data from tissue microarrays (TMAs) to identify six major cellular and acellular constituents of tissue, namely adipocytes, blood, collagen, epithelium, necrosis, and myofibroblasts. Experimental results show that the use of spatial information in addition to the spectral information brings significant improvements in the classifier performance and allows classification of cellular subtypes, such as adipocytes, that exhibit minimal chemical information but have distinct spatial characteristics. This work demonstrates the application and efficiency of deep learning algorithms in improving the diagnostic techniques in clinical and research activities related to cancer.

Original languageEnglish (US)
Pages (from-to)1642-1653
Number of pages12
JournalAnalyst
Volume144
Issue number5
DOIs
StatePublished - Mar 7 2019

Fingerprint

Histology
histology
Fourier Analysis
Fourier transform
Fourier transforms
learning
Learning
Tissue
Infrared radiation
Neural networks
Adipocytes
cancer
Classifiers
Aptitude
Myofibroblasts
Biopsy
collagen
Image classification
image classification
Microarrays

ASJC Scopus subject areas

  • Analytical Chemistry
  • Biochemistry
  • Environmental Chemistry
  • Spectroscopy
  • Electrochemistry

Cite this

Berisha, S., Lotfollahi, M., Jahanipour, J., Gurcan, I., Walsh, M., Bhargava, R., ... Mayerich, D. (2019). Deep learning for FTIR histology: leveraging spatial and spectral features with convolutional neural networks. Analyst, 144(5), 1642-1653. https://doi.org/10.1039/c8an01495g

Deep learning for FTIR histology : leveraging spatial and spectral features with convolutional neural networks. / Berisha, Sebastian; Lotfollahi, Mahsa; Jahanipour, Jahandar; Gurcan, Ilker; Walsh, Michael; Bhargava, Rohit; Van Nguyen, Hien; Mayerich, David.

In: Analyst, Vol. 144, No. 5, 07.03.2019, p. 1642-1653.

Research output: Contribution to journalArticle

Berisha, S, Lotfollahi, M, Jahanipour, J, Gurcan, I, Walsh, M, Bhargava, R, Van Nguyen, H & Mayerich, D 2019, 'Deep learning for FTIR histology: leveraging spatial and spectral features with convolutional neural networks', Analyst, vol. 144, no. 5, pp. 1642-1653. https://doi.org/10.1039/c8an01495g
Berisha, Sebastian ; Lotfollahi, Mahsa ; Jahanipour, Jahandar ; Gurcan, Ilker ; Walsh, Michael ; Bhargava, Rohit ; Van Nguyen, Hien ; Mayerich, David. / Deep learning for FTIR histology : leveraging spatial and spectral features with convolutional neural networks. In: Analyst. 2019 ; Vol. 144, No. 5. pp. 1642-1653.
@article{4b1d5631f99c42e08432a4590662d331,
title = "Deep learning for FTIR histology: leveraging spatial and spectral features with convolutional neural networks",
abstract = "Current methods for cancer detection rely on tissue biopsy, chemical labeling/staining, and examination of the tissue by a pathologist. Though these methods continue to remain the gold standard, they are non-quantitative and susceptible to human error. Fourier transform infrared (FTIR) spectroscopic imaging has shown potential as a quantitative alternative to traditional histology. However, identification of histological components requires reliable classification based on molecular spectra, which are susceptible to artifacts introduced by noise and scattering. Several tissue types, particularly in heterogeneous tissue regions, tend to confound traditional classification methods. Convolutional neural networks (CNNs) are the current state-of-the-art in image classification, providing the ability to learn spatial characteristics of images. In this paper, we demonstrate that CNNs with architectures designed to process both spectral and spatial information can significantly improve classifier performance over per-pixel spectral classification. We report classification results after applying CNNs to data from tissue microarrays (TMAs) to identify six major cellular and acellular constituents of tissue, namely adipocytes, blood, collagen, epithelium, necrosis, and myofibroblasts. Experimental results show that the use of spatial information in addition to the spectral information brings significant improvements in the classifier performance and allows classification of cellular subtypes, such as adipocytes, that exhibit minimal chemical information but have distinct spatial characteristics. This work demonstrates the application and efficiency of deep learning algorithms in improving the diagnostic techniques in clinical and research activities related to cancer.",
author = "Sebastian Berisha and Mahsa Lotfollahi and Jahandar Jahanipour and Ilker Gurcan and Michael Walsh and Rohit Bhargava and {Van Nguyen}, Hien and David Mayerich",
year = "2019",
month = "3",
day = "7",
doi = "10.1039/c8an01495g",
language = "English (US)",
volume = "144",
pages = "1642--1653",
journal = "Analyst",
issn = "0003-2654",
publisher = "Royal Society of Chemistry",
number = "5",

}

TY - JOUR

T1 - Deep learning for FTIR histology

T2 - leveraging spatial and spectral features with convolutional neural networks

AU - Berisha, Sebastian

AU - Lotfollahi, Mahsa

AU - Jahanipour, Jahandar

AU - Gurcan, Ilker

AU - Walsh, Michael

AU - Bhargava, Rohit

AU - Van Nguyen, Hien

AU - Mayerich, David

PY - 2019/3/7

Y1 - 2019/3/7

N2 - Current methods for cancer detection rely on tissue biopsy, chemical labeling/staining, and examination of the tissue by a pathologist. Though these methods continue to remain the gold standard, they are non-quantitative and susceptible to human error. Fourier transform infrared (FTIR) spectroscopic imaging has shown potential as a quantitative alternative to traditional histology. However, identification of histological components requires reliable classification based on molecular spectra, which are susceptible to artifacts introduced by noise and scattering. Several tissue types, particularly in heterogeneous tissue regions, tend to confound traditional classification methods. Convolutional neural networks (CNNs) are the current state-of-the-art in image classification, providing the ability to learn spatial characteristics of images. In this paper, we demonstrate that CNNs with architectures designed to process both spectral and spatial information can significantly improve classifier performance over per-pixel spectral classification. We report classification results after applying CNNs to data from tissue microarrays (TMAs) to identify six major cellular and acellular constituents of tissue, namely adipocytes, blood, collagen, epithelium, necrosis, and myofibroblasts. Experimental results show that the use of spatial information in addition to the spectral information brings significant improvements in the classifier performance and allows classification of cellular subtypes, such as adipocytes, that exhibit minimal chemical information but have distinct spatial characteristics. This work demonstrates the application and efficiency of deep learning algorithms in improving the diagnostic techniques in clinical and research activities related to cancer.

AB - Current methods for cancer detection rely on tissue biopsy, chemical labeling/staining, and examination of the tissue by a pathologist. Though these methods continue to remain the gold standard, they are non-quantitative and susceptible to human error. Fourier transform infrared (FTIR) spectroscopic imaging has shown potential as a quantitative alternative to traditional histology. However, identification of histological components requires reliable classification based on molecular spectra, which are susceptible to artifacts introduced by noise and scattering. Several tissue types, particularly in heterogeneous tissue regions, tend to confound traditional classification methods. Convolutional neural networks (CNNs) are the current state-of-the-art in image classification, providing the ability to learn spatial characteristics of images. In this paper, we demonstrate that CNNs with architectures designed to process both spectral and spatial information can significantly improve classifier performance over per-pixel spectral classification. We report classification results after applying CNNs to data from tissue microarrays (TMAs) to identify six major cellular and acellular constituents of tissue, namely adipocytes, blood, collagen, epithelium, necrosis, and myofibroblasts. Experimental results show that the use of spatial information in addition to the spectral information brings significant improvements in the classifier performance and allows classification of cellular subtypes, such as adipocytes, that exhibit minimal chemical information but have distinct spatial characteristics. This work demonstrates the application and efficiency of deep learning algorithms in improving the diagnostic techniques in clinical and research activities related to cancer.

UR - http://www.scopus.com/inward/record.url?scp=85062097377&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062097377&partnerID=8YFLogxK

U2 - 10.1039/c8an01495g

DO - 10.1039/c8an01495g

M3 - Article

VL - 144

SP - 1642

EP - 1653

JO - Analyst

JF - Analyst

SN - 0003-2654

IS - 5

ER -