Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach

Aerts, Hugo J.W.L.; Velazquez, Emmanuel Rios; Leijenaar, Ralph T.H.; Parmar, Chintan; Großmann, Patrick; Carvalho, Sara; Cavalho, Sara; Bussink, Johan; Monshouwer, R.; Haibe‐Kains, Benjamin; Rietveld, D.; Hoebers, Frank; Rietbergen, Michelle M.; Leemans, C. René; Dekker, André; Quackenbush, John; Gillies, Robert J.; Lambin, Philippe

doi:10.1038/ncomms5006

Cited by 3,858 publications

(3,834 citation statements)

References 32 publications

Supporting

Mentioning

3,576

Contrasting

Unclassified

Order By: Relevance

“…Also, they are formulated to have a maximum value when all elements in the image are of equal values. Therefore, these features are characterized by high sensitivity to the presence of adjacent diagonal elements in the GLCM 65, 66. These characteristics might lead to their remarkable insensitivity toward variation of the studied parameters.…”

Section: Discussionmentioning

confidence: 99%

Reproducibility of F18‐FDG PET radiomic features for different cervical tumor segmentation methods, gray‐level discretization, and reconstruction algorithms

Altazi

Zhang

Fernandez

et al. 2017

J Applied Clin Med Phys

View full text Add to dashboard Cite

Site‐specific investigations of the role of radiomics in cancer diagnosis and therapy are emerging. We evaluated the reproducibility of radiomic features extracted from 18Flourine–fluorodeoxyglucose (18F‐FDG) PET images for three parameters: manual versus computer‐aided segmentation methods, gray‐level discretization, and PET image reconstruction algorithms. Our cohort consisted of pretreatment PET/CT scans from 88 cervical cancer patients. Two board‐certified radiation oncologists manually segmented the metabolic tumor volume (MTV1 and MTV2) for each patient. For comparison, we used a graphical‐based method to generate semiautomated segmented volumes (GBSV). To address any perturbations in radiomic feature values, we down‐sampled the tumor volumes into three gray‐levels: 32, 64, and 128 from the original gray‐level of 256. Finally, we analyzed the effect on radiomic features on PET images of eight patients due to four PET 3D‐reconstruction algorithms: maximum likelihood‐ordered subset expectation maximization (OSEM) iterative reconstruction (IR) method, fourier rebinning‐ML‐OSEM (FOREIR), FORE‐filtered back projection (FOREFBP), and 3D‐Reprojection (3DRP) analytical method. We extracted 79 features from all segmentation method, gray‐levels of down‐sampled volumes, and PET reconstruction algorithms. The features were extracted using gray‐level co‐occurrence matrices (GLCM), gray‐level size zone matrices (GLSZM), gray‐level run‐length matrices (GLRLM), neighborhood gray‐tone difference matrices (NGTDM), shape‐based features (SF), and intensity histogram features (IHF). We computed the Dice coefficient between each MTV and GBSV to measure segmentation accuracy. Coefficient values close to one indicate high agreement, and values close to zero indicate low agreement. We evaluated the effect on radiomic features by calculating the mean percentage differences (d¯) between feature values measured from each pair of parameter elements (i.e. segmentation methods: MTV1‐MTV2, MTV1‐GBSV, MTV2‐GBSV; gray‐levels: 64‐32, 64‐128, and 64‐256; reconstruction algorithms: OSEM‐FORE‐OSEM, OSEM‐FOREFBP, and OSEM‐3DRP). We used false|normald¯false| as a measure of radiomic feature reproducibility level, where any feature scored false|normald¯false| ±SD ≤ |25|% ± 35% was considered reproducible. We used Bland–Altman analysis to evaluate the mean, standard deviation (SD), and upper/lower reproducibility limits (U/LRL) for radiomic features in response to variation in each testing parameter. Furthermore, we proposed U/LRL as a method to classify the level of reproducibility: High— ±1% ≤ U/LRL ≤ ±30%; Intermediate— ±30% < U/LRL ≤ ±45%; Low— ±45 < U/LRL ≤ ±50%. We considered any feature below the low level as nonreproducible (NR). Finally, we calculated the interclass correlation coefficient (ICC) to evaluate the reliability of radiomic feature measurements for each parameter. The segmented volumes of 65 patients (81.3%) scored Dice coefficient >0.75 for all three volumes. The result outcomes revealed a tendency of higher radiomic fe...

show abstract

Section: Discussionmentioning

confidence: 99%

Reproducibility of F18‐FDG PET radiomic features for different cervical tumor segmentation methods, gray‐level discretization, and reconstruction algorithms

Altazi

Zhang

Fernandez

et al. 2017

J Applied Clin Med Phys

View full text Add to dashboard Cite

show abstract

“…In total, 68 radiomic features were extracted from the PET images, including 7 shape features (29), 13 histogram-based features (29), and 48 texture features. The texture features included 22 gray level cooccurrence matrix (32), 11 run length matrix (33), 10 size zone matrix (34), and 5 neighborhood gray-tone difference matrix (35) features (Supplemental Table 1; supplemental materials are available at http:// jnm.snmjournals.org).…”

Section: Pet Feature Extraction and Selectionmentioning

confidence: 99%

“…Therefore, accurate quantification of tumor heterogeneity from PET images may provide important information for the identification of mutation status and precision medicine. Heterogeneity in the tumor phenotype can be quantitatively described through radiomic features (23,27,28), which use advanced mathematic models to quantify the spatial relationship between image voxels (29).…”

mentioning

confidence: 99%

Associations Between Somatic Mutations and Metabolic Imaging Phenotypes in Non–Small Cell Lung Cancer

et al. 2016

Self Cite

View full text Add to dashboard Cite

PET-based radiomics have been used to noninvasively quantify the metabolic tumor phenotypes; however, little is known about the relationship between these phenotypes and underlying somatic mutations. This study assessed the association and predictive power of 18 F-FDG PET-based radiomic features for somatic mutations in non-small cell lung cancer patients. Methods: Three hundred forty-eight non-small cell lung cancer patients underwent diagnostic 18 F-FDG PET scans and were tested for genetic mutations. Thirteen percent (44/348) and 28% (96/348) of patients were found to harbor epidermal growth factor receptor (EGFR) or Kristen rat sarcoma viral (KRAS) mutations, respectively. We evaluated 21 imaging features: 19 independent radiomic features quantifying phenotypic traits and 2 conventional features (metabolic tumor volume and maximum SUV). The association between imaging features and mutation status (e.g., EGFR-positive [EGFR1] vs. EGFR-negative) was assessed using the Wilcoxon rank-sum test. The ability of each imaging feature to predict mutation status was evaluated by the area under the receiver operating curve (AUC) and its significance was compared with a random guess (AUC 5 0.5) using the Noether test. All P values were corrected for multiple hypothesis testing by controlling the false-discovery rate (FDR Wilcoxon , FDR Noether ) with a significance threshold of 10%. Results: Eight radiomic features and both conventional features were significantly associated with EGFR mutation status (FDR Wilcoxon 5 0.01-0.10). One radiomic feature (normalized inverse difference moment) outperformed all other features in predicting EGFR mutation status (EGFR1 vs. EGFR-negative, AUC 5 0.67, FDR Noether 5 0.0032), as well as differentiating between KRAS-positive and EGFR1 (AUC 5 0.65, FDR Noether 5 0.05). None of the features was associated with or predictive of KRAS mutation status (KRAS-positive vs. KRAS-negative, AUC 5 0.50-0.54). Conclusion: Our results indicate that EGFR mutations may drive different metabolic tumor phenotypes that are captured in PET images, whereas KRAS-mutated tumors do not. This proof-of-concept study sheds light on genotype-phenotype interactions, using radiomics to capture and describe the phenotype, and may have potential for developing noninvasive imaging biomarkers for somatic mutations.

show abstract

“…11,12 This emerging field known as radiomics 13 has been facilitated by large computing power aimed at finding quantitative features in support of medical decisions. 14 This development has, however and once more, questioned the role of human readers in future medical devices. For some tasks, computational methods are achieving performance levels approaching those of experienced clinicians and higher than those attained by poorly trained practitioners.…”

mentioning

confidence: 99%

“How much realism is needed?” — the wrong question in silico imagers have been asking

Badano

2017

Medical Physics

View full text Add to dashboard Cite

Purpose: To discuss the use of realism as a first approximation for assessing computational imaging methods. Methods: Although in silico methods are increasingly becoming promising surrogates to physical experimentation for various stages of device development, their acceptance remains challenging. Realism is often considered as a first approximation for assessing computational imaging methods. However, realism is subjective and does not always ensure that key features of the methodologies reflect relevant aspects of devices of interest to imaging scientists, regulators, and medical practitioners. Moreover, in some cases (e.g., in computerized image analysis applications where human interpretation is not needed) how realistic in silico images are is irrelevant and perhaps misleading. Results: I emphasize a divergence from this methodology by providing a rationale for evaluating in silico imaging methods and tools in an objective and measurable manner. Conclusions: Improved approaches for in silico imaging will lead to the rapid advancement and acceptance of computational techniques in medical imaging primarily but not limited to the regulatory evaluation of new imaging products. Published 2017. This article is a U.S. Government work and is in the public domain in the USA. The realism of computer-generated imagery as simulation of real-world images is often used as part of the development and testing of new in silico imaging methods. Many imaging scientists use realism of the synthetic images as a key success indicator for an in silico methodology. Typical studies that investigate realism use a human interrogator (image reader) to visually evaluate images using a discrete scale with a blind experimental paradigm. In this type of experiment, sometimes known as a "fool the radiologist" study design, the experiment is deemed successful if the performance of the radiologists is found to be at almost guessing levels for distinguishing synthetic versus real-world images. A typical example of this procedure is found in Shaheen et al. 1 where synthetic mammographic lesions were inserted into normal patient images to augment datasets for training and testing of image analysis software. The study consisted of inserting MRI-derived, spiculated lesions into projection images to obtain mammograms and into tomographic sequences of projection images reconstructed to form tomosynthesis datasets. The realism of these models was assessed by five radiologists using a receiver operating characteristic (ROC) study as compared to 54 real-world masses. The authors reported that humans were only marginally capable of distinguishing synthetic from real-world images with area-under-the-ROC curve averaged over all observers of 0.54 [95% confidence interval (0.50, 0.66)] for the 2D study and 0.67 [95% confidence interval (0.55, 0.79)] for tomosynthesis. The study concluded that the mass models were validated for the realism of their appearance for digital mammography and breast tomosynthesis and proposed the database of models as suitable f...

show abstract

Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach

Cited by 3,858 publications

References 32 publications

Reproducibility of F18‐FDG PET radiomic features for different cervical tumor segmentation methods, gray‐level discretization, and reconstruction algorithms

Reproducibility of F18‐FDG PET radiomic features for different cervical tumor segmentation methods, gray‐level discretization, and reconstruction algorithms

Associations Between Somatic Mutations and Metabolic Imaging Phenotypes in Non–Small Cell Lung Cancer

“How much realism is needed?” — the wrong question in silico imagers have been asking

Contact Info

Product

Resources

About