Exploratory study to identify radiomics classifiers for lung cancer histology
Background: Radiomics can quantify tumor phenotypic characteristics non-invasively by applying feature algorithms to medical imaging data. In this study of lung cancer patients, we investigated the association between radiomic features and the tumor histologic subtypes (adenocarcinoma and squamous c...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2016-03-01
|
| Series: | Frontiers in Oncology |
| Subjects: | |
| Online Access: | http://journal.frontiersin.org/Journal/10.3389/fonc.2016.00071/full |
| _version_ | 1828329604803198976 |
|---|---|
| author | Weimiao eWu Weimiao eWu Chintan eParmar Chintan eParmar Chintan eParmar Patrick eGrossmann Patrick eGrossmann John eQuackenbush John eQuackenbush Philippe eLambin Johan eBussink Raymond eMak Hugo eAerts Hugo eAerts |
| author_facet | Weimiao eWu Weimiao eWu Chintan eParmar Chintan eParmar Chintan eParmar Patrick eGrossmann Patrick eGrossmann John eQuackenbush John eQuackenbush Philippe eLambin Johan eBussink Raymond eMak Hugo eAerts Hugo eAerts |
| author_sort | Weimiao eWu |
| collection | DOAJ |
| description | Background: Radiomics can quantify tumor phenotypic characteristics non-invasively by applying feature algorithms to medical imaging data. In this study of lung cancer patients, we investigated the association between radiomic features and the tumor histologic subtypes (adenocarcinoma and squamous cell carcinoma). Furthermore, in order to predict histologic subtypes, we employed machine-learning methods and independently evaluated their prediction performance. Methods: Two independent radiomic cohorts with a combined size of 350 patients were included in our analysis. A total of 440 radiomic features were extracted from the segmented tumor volumes of pretreatment CT images. These radiomic features quantify tumor phenotypic characteristics on medical images using tumor shape and size, intensity statistics and texture. Univariate analysis was performed to assess each feature’s association with the histological subtypes. In our multivariate analysis, we investigated 24 feature selection methods and three classification methods for histology prediction. Multivariate models were trained on the training cohort and their performance was evaluated on the independent validation cohort using the area under ROC curve (AUC). Histology was determined from surgical specimen. Results: In our univariate analysis, we observed that fifty-three radiomic features were significantly associated with tumor histology. Moreover, sixteen of these fifty-three features showed significant differences across the two histological subtypes according to the KS test in validation dataset. In multivariate analysis, feature selection methods ReliefF and its variants showed higher prediction accuracy as compared to other methods. We found that Naive Bayes classifier outperforms other classifiers and achieved the highest AUC (0.72; p-value = 2.3×〖10〗^(-7)) with five features: Stats_min, Wavelet_HLL_rlgl_lowGrayLevelRunEmphasis, Wavelet_HHL_stats_median, Wavelet_HLL_stats_skewness, Wavelet_HLH_glcm_clusShade. Conclusions: Histological subtypes can influence the choice of a treatment/therapy for lung cancer patients. We observed that radiomic features show significant association with the lung tumor histology. Moreover, radiomics based multivariate classifiers were independently validated for the prediction of histological subtypes. Despite achieving lower than optimal prediction accuracy (AUC 0.72), our analysis highlights the impressive potential of non-invasive and cost-effective radiomics for precision medicine. Further research in this direction could lead us to optimal performance and therefore to clinical applicability, which could enhance the efficiency and efficacy of cancer care. |
| first_indexed | 2024-04-13T20:27:57Z |
| format | Article |
| id | doaj.art-a93549026f0d480cab56952e84317e90 |
| institution | Directory Open Access Journal |
| issn | 2234-943X |
| language | English |
| last_indexed | 2024-04-13T20:27:57Z |
| publishDate | 2016-03-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Oncology |
| spelling | doaj.art-a93549026f0d480cab56952e84317e902022-12-22T02:31:16ZengFrontiers Media S.A.Frontiers in Oncology2234-943X2016-03-01610.3389/fonc.2016.00071175984Exploratory study to identify radiomics classifiers for lung cancer histologyWeimiao eWu0Weimiao eWu1Chintan eParmar2Chintan eParmar3Chintan eParmar4Patrick eGrossmann5Patrick eGrossmann6John eQuackenbush7John eQuackenbush8Philippe eLambin9Johan eBussink10Raymond eMak11Hugo eAerts12Hugo eAerts13Harvard T.H. Chan School of Public HealthDana-Farber Cancer InstituteDana-Farber Cancer InstituteBrigham and Women’s Hospital, Harvard Medical SchoolMaastricht UniversityDana-Farber Cancer InstituteBrigham and Women’s Hospital, Harvard Medical SchoolHarvard T.H. Chan School of Public HealthDana-Farber Cancer InstituteMaastricht UniversityRadboud university medical centerDana-Farber Cancer InstituteDana-Farber Cancer InstituteBrigham and Women’s Hospital, Harvard Medical SchoolBackground: Radiomics can quantify tumor phenotypic characteristics non-invasively by applying feature algorithms to medical imaging data. In this study of lung cancer patients, we investigated the association between radiomic features and the tumor histologic subtypes (adenocarcinoma and squamous cell carcinoma). Furthermore, in order to predict histologic subtypes, we employed machine-learning methods and independently evaluated their prediction performance. Methods: Two independent radiomic cohorts with a combined size of 350 patients were included in our analysis. A total of 440 radiomic features were extracted from the segmented tumor volumes of pretreatment CT images. These radiomic features quantify tumor phenotypic characteristics on medical images using tumor shape and size, intensity statistics and texture. Univariate analysis was performed to assess each feature’s association with the histological subtypes. In our multivariate analysis, we investigated 24 feature selection methods and three classification methods for histology prediction. Multivariate models were trained on the training cohort and their performance was evaluated on the independent validation cohort using the area under ROC curve (AUC). Histology was determined from surgical specimen. Results: In our univariate analysis, we observed that fifty-three radiomic features were significantly associated with tumor histology. Moreover, sixteen of these fifty-three features showed significant differences across the two histological subtypes according to the KS test in validation dataset. In multivariate analysis, feature selection methods ReliefF and its variants showed higher prediction accuracy as compared to other methods. We found that Naive Bayes classifier outperforms other classifiers and achieved the highest AUC (0.72; p-value = 2.3×〖10〗^(-7)) with five features: Stats_min, Wavelet_HLL_rlgl_lowGrayLevelRunEmphasis, Wavelet_HHL_stats_median, Wavelet_HLL_stats_skewness, Wavelet_HLH_glcm_clusShade. Conclusions: Histological subtypes can influence the choice of a treatment/therapy for lung cancer patients. We observed that radiomic features show significant association with the lung tumor histology. Moreover, radiomics based multivariate classifiers were independently validated for the prediction of histological subtypes. Despite achieving lower than optimal prediction accuracy (AUC 0.72), our analysis highlights the impressive potential of non-invasive and cost-effective radiomics for precision medicine. Further research in this direction could lead us to optimal performance and therefore to clinical applicability, which could enhance the efficiency and efficacy of cancer care.http://journal.frontiersin.org/Journal/10.3389/fonc.2016.00071/fullcomputational scienceFeature Selectionquantitative imagingLung cancer histologyRadiomics |
| spellingShingle | Weimiao eWu Weimiao eWu Chintan eParmar Chintan eParmar Chintan eParmar Patrick eGrossmann Patrick eGrossmann John eQuackenbush John eQuackenbush Philippe eLambin Johan eBussink Raymond eMak Hugo eAerts Hugo eAerts Exploratory study to identify radiomics classifiers for lung cancer histology Frontiers in Oncology computational science Feature Selection quantitative imaging Lung cancer histology Radiomics |
| title | Exploratory study to identify radiomics classifiers for lung cancer histology |
| title_full | Exploratory study to identify radiomics classifiers for lung cancer histology |
| title_fullStr | Exploratory study to identify radiomics classifiers for lung cancer histology |
| title_full_unstemmed | Exploratory study to identify radiomics classifiers for lung cancer histology |
| title_short | Exploratory study to identify radiomics classifiers for lung cancer histology |
| title_sort | exploratory study to identify radiomics classifiers for lung cancer histology |
| topic | computational science Feature Selection quantitative imaging Lung cancer histology Radiomics |
| url | http://journal.frontiersin.org/Journal/10.3389/fonc.2016.00071/full |
| work_keys_str_mv | AT weimiaoewu exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT weimiaoewu exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT chintaneparmar exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT chintaneparmar exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT chintaneparmar exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT patrickegrossmann exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT patrickegrossmann exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT johnequackenbush exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT johnequackenbush exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT philippeelambin exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT johanebussink exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT raymondemak exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT hugoeaerts exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology AT hugoeaerts exploratorystudytoidentifyradiomicsclassifiersforlungcancerhistology |