Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images

Duc Fehr, Harini Veeraraghavan, Andreas Wibmer, Tatsuo Gondo, Kazuhiro Matsumoto, Herbert Alberto Vargas, Evis Sala, Hedvig Hricak, Joseph O. Deasy

Research output: Contribution to journalArticle

105 Citations (Scopus)

Abstract

Noninvasive, radiological image-based detection and stratification of Gleason patterns can impact clinical outcomes, treatment selection, and the determination of disease status at diagnosis without subjecting patients to surgical biopsies. We present machine learning-based automatic classification of prostate cancer aggressiveness by combining apparent diffusion coefficient (ADC) and T2-weighted (T2-w) MRI-based texture features. Our approach achieved reasonably accurate classification of Gleason scores (GS) 6(3 + 3) vs. ≥7 and 7(3 + 4) vs. 7(4 + 3) despite the presence of highly unbalanced samples by using two different sample augmentation techniques followed by feature selection-based classification. Our method distinguished between GS 6(3 + 3) and ≥7 cancers with 93% accuracy for cancers occurring in both peripheral (PZ) and transition (TZ) zones and 92% for cancers occurring in the PZ alone. Our approach distinguished the GS 7(3 + 4) from GS 7(4 + 3) with 92% accuracy for cancers occurring in both the PZ and TZ and with 93% for cancers occurring in the PZ alone. In comparison, a classifier using only the ADC mean achieved a top accuracy of 58% for distinguishing GS 6(3 + 3) vs. GS ≥7 for cancers occurring in PZ and TZ and 63% for cancers occurring in PZ alone. The same classifier achieved an accuracy of 59% for distinguishing GS 7(3 + 4) from GS 7(4 + 3) occurring in the PZ and TZ and 60% for cancers occurring in PZ alone. Separate analysis of the cancers occurring in TZ alone was not performed owing to the limited number of samples. Our results suggest that texture features derived from ADC and T2-w MRI together with sample augmentation can help to obtain reasonably accurate classification of Gleason patterns.

Original languageEnglish
Pages (from-to)E6265-E6273
JournalProceedings of the National Academy of Sciences of the United States of America
Volume112
Issue number46
DOIs
Publication statusPublished - 2015 Nov 17
Externally publishedYes

Fingerprint

Neoplasm Grading
Prostatic Neoplasms
Magnetic Resonance Spectroscopy
Neoplasms
Biopsy

Keywords

  • Gleason score classification
  • Learning from unbalanced data
  • Multiparametric mri
  • PCa gleason (3+4) vs. (4+3) cancers
  • PCa gleason 6 vs. ≥7

ASJC Scopus subject areas

  • General

Cite this

Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images. / Fehr, Duc; Veeraraghavan, Harini; Wibmer, Andreas; Gondo, Tatsuo; Matsumoto, Kazuhiro; Vargas, Herbert Alberto; Sala, Evis; Hricak, Hedvig; Deasy, Joseph O.

In: Proceedings of the National Academy of Sciences of the United States of America, Vol. 112, No. 46, 17.11.2015, p. E6265-E6273.

Research output: Contribution to journalArticle

Fehr, Duc ; Veeraraghavan, Harini ; Wibmer, Andreas ; Gondo, Tatsuo ; Matsumoto, Kazuhiro ; Vargas, Herbert Alberto ; Sala, Evis ; Hricak, Hedvig ; Deasy, Joseph O. / Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images. In: Proceedings of the National Academy of Sciences of the United States of America. 2015 ; Vol. 112, No. 46. pp. E6265-E6273.
@article{03c086f8f24b41189ced3e461ceb9776,
title = "Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images",
abstract = "Noninvasive, radiological image-based detection and stratification of Gleason patterns can impact clinical outcomes, treatment selection, and the determination of disease status at diagnosis without subjecting patients to surgical biopsies. We present machine learning-based automatic classification of prostate cancer aggressiveness by combining apparent diffusion coefficient (ADC) and T2-weighted (T2-w) MRI-based texture features. Our approach achieved reasonably accurate classification of Gleason scores (GS) 6(3 + 3) vs. ≥7 and 7(3 + 4) vs. 7(4 + 3) despite the presence of highly unbalanced samples by using two different sample augmentation techniques followed by feature selection-based classification. Our method distinguished between GS 6(3 + 3) and ≥7 cancers with 93{\%} accuracy for cancers occurring in both peripheral (PZ) and transition (TZ) zones and 92{\%} for cancers occurring in the PZ alone. Our approach distinguished the GS 7(3 + 4) from GS 7(4 + 3) with 92{\%} accuracy for cancers occurring in both the PZ and TZ and with 93{\%} for cancers occurring in the PZ alone. In comparison, a classifier using only the ADC mean achieved a top accuracy of 58{\%} for distinguishing GS 6(3 + 3) vs. GS ≥7 for cancers occurring in PZ and TZ and 63{\%} for cancers occurring in PZ alone. The same classifier achieved an accuracy of 59{\%} for distinguishing GS 7(3 + 4) from GS 7(4 + 3) occurring in the PZ and TZ and 60{\%} for cancers occurring in PZ alone. Separate analysis of the cancers occurring in TZ alone was not performed owing to the limited number of samples. Our results suggest that texture features derived from ADC and T2-w MRI together with sample augmentation can help to obtain reasonably accurate classification of Gleason patterns.",
keywords = "Gleason score classification, Learning from unbalanced data, Multiparametric mri, PCa gleason (3+4) vs. (4+3) cancers, PCa gleason 6 vs. ≥7",
author = "Duc Fehr and Harini Veeraraghavan and Andreas Wibmer and Tatsuo Gondo and Kazuhiro Matsumoto and Vargas, {Herbert Alberto} and Evis Sala and Hedvig Hricak and Deasy, {Joseph O.}",
year = "2015",
month = "11",
day = "17",
doi = "10.1073/pnas.1505935112",
language = "English",
volume = "112",
pages = "E6265--E6273",
journal = "Proceedings of the National Academy of Sciences of the United States of America",
issn = "0027-8424",
number = "46",

}

TY - JOUR

T1 - Automatic classification of prostate cancer Gleason scores from multiparametric magnetic resonance images

AU - Fehr, Duc

AU - Veeraraghavan, Harini

AU - Wibmer, Andreas

AU - Gondo, Tatsuo

AU - Matsumoto, Kazuhiro

AU - Vargas, Herbert Alberto

AU - Sala, Evis

AU - Hricak, Hedvig

AU - Deasy, Joseph O.

PY - 2015/11/17

Y1 - 2015/11/17

N2 - Noninvasive, radiological image-based detection and stratification of Gleason patterns can impact clinical outcomes, treatment selection, and the determination of disease status at diagnosis without subjecting patients to surgical biopsies. We present machine learning-based automatic classification of prostate cancer aggressiveness by combining apparent diffusion coefficient (ADC) and T2-weighted (T2-w) MRI-based texture features. Our approach achieved reasonably accurate classification of Gleason scores (GS) 6(3 + 3) vs. ≥7 and 7(3 + 4) vs. 7(4 + 3) despite the presence of highly unbalanced samples by using two different sample augmentation techniques followed by feature selection-based classification. Our method distinguished between GS 6(3 + 3) and ≥7 cancers with 93% accuracy for cancers occurring in both peripheral (PZ) and transition (TZ) zones and 92% for cancers occurring in the PZ alone. Our approach distinguished the GS 7(3 + 4) from GS 7(4 + 3) with 92% accuracy for cancers occurring in both the PZ and TZ and with 93% for cancers occurring in the PZ alone. In comparison, a classifier using only the ADC mean achieved a top accuracy of 58% for distinguishing GS 6(3 + 3) vs. GS ≥7 for cancers occurring in PZ and TZ and 63% for cancers occurring in PZ alone. The same classifier achieved an accuracy of 59% for distinguishing GS 7(3 + 4) from GS 7(4 + 3) occurring in the PZ and TZ and 60% for cancers occurring in PZ alone. Separate analysis of the cancers occurring in TZ alone was not performed owing to the limited number of samples. Our results suggest that texture features derived from ADC and T2-w MRI together with sample augmentation can help to obtain reasonably accurate classification of Gleason patterns.

AB - Noninvasive, radiological image-based detection and stratification of Gleason patterns can impact clinical outcomes, treatment selection, and the determination of disease status at diagnosis without subjecting patients to surgical biopsies. We present machine learning-based automatic classification of prostate cancer aggressiveness by combining apparent diffusion coefficient (ADC) and T2-weighted (T2-w) MRI-based texture features. Our approach achieved reasonably accurate classification of Gleason scores (GS) 6(3 + 3) vs. ≥7 and 7(3 + 4) vs. 7(4 + 3) despite the presence of highly unbalanced samples by using two different sample augmentation techniques followed by feature selection-based classification. Our method distinguished between GS 6(3 + 3) and ≥7 cancers with 93% accuracy for cancers occurring in both peripheral (PZ) and transition (TZ) zones and 92% for cancers occurring in the PZ alone. Our approach distinguished the GS 7(3 + 4) from GS 7(4 + 3) with 92% accuracy for cancers occurring in both the PZ and TZ and with 93% for cancers occurring in the PZ alone. In comparison, a classifier using only the ADC mean achieved a top accuracy of 58% for distinguishing GS 6(3 + 3) vs. GS ≥7 for cancers occurring in PZ and TZ and 63% for cancers occurring in PZ alone. The same classifier achieved an accuracy of 59% for distinguishing GS 7(3 + 4) from GS 7(4 + 3) occurring in the PZ and TZ and 60% for cancers occurring in PZ alone. Separate analysis of the cancers occurring in TZ alone was not performed owing to the limited number of samples. Our results suggest that texture features derived from ADC and T2-w MRI together with sample augmentation can help to obtain reasonably accurate classification of Gleason patterns.

KW - Gleason score classification

KW - Learning from unbalanced data

KW - Multiparametric mri

KW - PCa gleason (3+4) vs. (4+3) cancers

KW - PCa gleason 6 vs. ≥7

UR - http://www.scopus.com/inward/record.url?scp=84966335551&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84966335551&partnerID=8YFLogxK

U2 - 10.1073/pnas.1505935112

DO - 10.1073/pnas.1505935112

M3 - Article

VL - 112

SP - E6265-E6273

JO - Proceedings of the National Academy of Sciences of the United States of America

JF - Proceedings of the National Academy of Sciences of the United States of America

SN - 0027-8424

IS - 46

ER -