Could normalization improve robustness of abdominal MRI radiomic features?

Radiomics-based systems could improve the management of oncological patients by supporting cancer diagnosis, treatment planning, and response assessment. However, one of the main limitations of these systems is the generalizability and reproducibility of results when they are applied to images acquired in different hospitals by different scanners. Normalization has been introduced to mitigate this issue, and two main approaches have been proposed: one rescales the image intensities (image normalization), the other the feature distributions for each center (feature normalization). The aim of this study is to evaluate how different image and feature normalization methods impact the robustness of 93 radiomics features acquired using a multicenter and multi-scanner abdominal Magnetic Resonance Imaging (MRI) dataset. To this scope, 88 rectal MRIs were retrospectively collected from 3 different institutions (4 scanners), and for each patient, six 3D regions of interest on the obturator muscle were considered. The methods applied were min-max, 1st-99th percentiles and 3-Sigma normalization, z-score standardization, mean centering, histogram normalization, Nyul-Udupa and ComBat harmonization. The Mann-Whitney U-test was applied to assess features repeatability between scanners, by comparing the feature values obtained for each normalization method, including the case in which no normalization was applied. Most image normalization methods allowed to reduce the overall variability in terms of intensity distributions, while worsening or showing unpredictable results in terms of feature robustness, except for thez-score, which provided a slight improvement by increasing the number of statistically similar features from 9/93 to 10/93. Conversely, feature normalization methods positively reduced the overall variability across the scanners, in particular, 3sigma,z_scoreandComBatthat increased the number of similar features (79/93). According to our results, it emerged that none of the image normalization methods was able to strongly increase the number of statistically similar features.

Virtual biopsy in abdominal pathology: where do we stand?

In recent years, researchers have explored new ways to obtain information from pathological tissues, also exploring non-invasive techniques, such as virtual biopsy (VB). VB can be defined as a test that provides promising outcomes compared to traditional biopsy by extracting quantitative information from radiological images not accessible through traditional visual inspection. Data are processed in such a way that they can be correlated with the patient's phenotypic expression, or with molecular patterns and mutations, creating a bridge between traditional radiology, pathology, genomics, and artificial intelligence (AI). Radiomics is the backbone of VB, since it allows the extraction and selection of features from radiological images, feeding them into AI models in order to derive lesions' pathological characteristics and molecular status. Presently, the output of VB provides only a gross approximation of the findings of tissue biopsy. However, in the future, with the improvement of imaging resolution and processing techniques, VB could partially substitute the classical surgical or percutaneous biopsy, with the advantage of being non-invasive, comprehensive, accounting for lesion heterogeneity, and low cost. In this review, we investigate the concept of VB in abdominal pathology, focusing on its pipeline development and potential benefits.

A fully automatic deep learning algorithm to segment rectal Cancer on MR images: a multi-center study.

The aim of the study is to present and tune a fully automatic deep learning algorithm to segment colorectal cancers (CRC) on MR images, based on a U-Net structure. It is a multicenter study, including 3 different Italian institutions, that used 4 different MRI scanners. Two of them were used for training and tuning the systems, while the other two for the validation. The implemented algorithm consists of a pre-processing step to normalize and to highlight the tumoral area, followed by the CRC segmentation using different U-net structures. Automatic masks were compared with manual segmentations performed by three experienced radiologists, one at each center. The two best performing systems (called mdl2 and mdl3), obtained a median Dice Similarity Coefficient of 0.68(mdl2) - 0.69(mdl3), precision of 0.75(md/2) - 0.71(md/3), and recall of 0.69(mdl2) - 0.73(mdl3) on the validation set. Both systems reached high detection rates, 0.98 and 0.95, respectively, on the validation set. These encouraging results, if confirmed on larger dataset, might improve the management of patients with CRC, since it can be used as a fast and precise tool for further radiomics analyses. Clinical Relevance - To provide a reliable tool able to automatically segment CRC tumors that can be used as first step in future radiomics studies aimed at predicting response to chemotherapy and personalizing treatment.

MRI-based radiomics to predict response in locally advanced rectal cancer: comparison of manual and automatic segmentation on external validation in a multicentre study.

BACKGROUND: Pathological complete response after neoadjuvant chemoradiotherapy in locally advanced rectal cancer (LARC) is achieved in 15-30% of cases. Our aim was to implement and externally validate a magnetic resonance imaging (MRI)-based radiomics pipeline to predict response to treatment and to investigate the impact of manual and automatic segmentations on the radiomics models. METHODS: Ninety-five patients with stage II/III LARC who underwent multiparametric MRI before chemoradiotherapy and surgical treatment were enrolled from three institutions. Patients were classified as responders if tumour regression grade was 1 or 2 and nonresponders otherwise. Sixty-seven patients composed the construction dataset, while 28 the external validation. Tumour volumes were manually and automatically segmented using a U-net algorithm. Three approaches for feature selection were tested and combined with four machine learning classifiers. RESULTS: Using manual segmentation, the best result reached an accuracy of 68% on the validation set, with sensitivity 60%, specificity 77%, negative predictive value (NPV) 63%, and positive predictive value (PPV) 75%. The automatic segmentation achieved an accuracy of 75% on the validation set, with sensitivity 80%, specificity 69%, and both NPV and PPV 75%. Sensitivity and NPV on the validation set were significantly higher (p = 0.047) for the automatic versus manual segmentation. CONCLUSION: Our study showed that radiomics models can pave the way to help clinicians in the prediction of tumour response to chemoradiotherapy of LARC and to personalise per-patient treatment. The results from the external validation dataset are promising for further research into radiomics approaches using both manual and automatic segmentations.

Comparison of radiomics approaches to predict resistance to 1st line chemotherapy in liver metastatic colorectal cancer.

Colorectal cancer (CRC) has the second-highest tumor incidence and is a leading cause of death by cancer. Nearly 20% of patients with CRC will have metastases (mts) at the time of diagnosis, and more than 50% of patients with CRC develop metastases during their disease. Unfortunately, only 45% of patients after a chemotherapy will respond to treatment. The aim of this study is to develop and validate a machine learning algorithm to predict response of individual liver mts, using CT scans. Understanding which mts will respond or not will help clinicians in providing a more efficient per-lesion treatment based on patient specific response and not only following a standard treatment. A group of 92 patients was enrolled from two Italian institutions. CT scans were collected, and the portal venous phase was manually segmented by an expert radiologist. Then, 75 radiomics features were extracted both from 7x7 ROIs that moved across the image and from the whole 3D mts. Feature selection was performed using a genetic algorithm. Results are presented as a comparison of the two different approaches of features extraction and different classification algorithms. Accuracy (ACC), sensitivity (SE), specificity (SP), negative and positive predictive values (NPV and PPV) were evaluated for all lesions (per-lesion analysis) and patients (per-patient analysis) in the construction and validation sets. Best results were obtained in the per-lesion analysis from the 3D approach using a Support Vector Machine as classifier. We reached on the training set an ACC of 81%, while on test set, we obtained SE of 76%, SP of 67%, PPV of 69% and NPV of 75%. On the validation set a SE of 61%, SP of 60%, PPV of 57% and NPV of 64% were reached. The promising results obtained in the validation dataset should be extended to a larger cohort of patient to further validate our method.Clinical Relevance- to develop a radiomics signatures predicting single liver mts response to therapy. A personalized mts approach is important to avoid unnecessary toxicity offering more suitable treatments and a better quality of life to oncological patients.

Deep learning model for automatic prostate segmentation on bicentric T2w images with and without endorectal coil.

Automatic segmentation of the prostate on Magnetic Resonance Imaging (MRI) is one of the topics on which research has focused in recent years as it is a fundamental first step in the building process of a Computer aided diagnosis (CAD) system for cancer detection. Unfortunately, MRI acquired in different centers with different scanners leads to images with different characteristics. In this work, we propose an automatic algorithm for prostate segmentation, based on a U-Net applying transfer learning method in a bi-center setting. First, T2w images with and without endorectal coil from 80 patients acquired at Center A were used as training set and internal validation set. Then, T2w images without endorectal coil from 20 patients acquired at Center B were used as external validation. The reference standard for this study was manual segmentation of the prostate gland performed by an expert operator. The results showed a Dice similarity coefficient >85% in both internal and external validation datasets.Clinical Relevance- This segmentation algorithm could be integrated into a CAD system to optimize computational effort in prostate cancer detection.

Radiomics and Magnetic Resonance Imaging of Rectal Cancer: From Engineering to Clinical Practice.

While cross-sectional imaging has seen continuous progress and plays an undiscussed pivotal role in the diagnostic management and treatment planning of patients with rectal cancer, a largely unmet need remains for improved staging accuracy, assessment of treatment response and prediction of individual patient outcome. Moreover, the increasing availability of target therapies has called for developing reliable diagnostic tools for identifying potential responders and optimizing overall treatment strategy on a personalized basis. Radiomics has emerged as a promising, still fully evolving research topic, which could harness the power of modern computer technology to generate quantitative information from imaging datasets based on advanced data-driven biomathematical models, potentially providing an added value to conventional imaging for improved patient management. The present study aimed to illustrate the contribution that current radiomics methods applied to magnetic resonance imaging can offer to managing patients with rectal cancer.

An innovative radiomics approach to predict response to chemotherapy of liver metastases based on CT images.

Liver metastases (mts) from colorectal cancer (CRC) can have different responses to chemotherapy in the same patient. The aim of this study is to develop and validate a machine learning algorithm to predict response of individual liver mts. 22 radiomic features (RF) were computed on pretreatment portal CT scans following a manual segmentation of mts. RFs were extracted from 7x7 Region of Interests (ROIs) that moved across the image by step of 2 pixels. Liver mts were classified as non-responder (R-) if their largest diameter increased more than 3 mm after 3 months of treatment and responder (R+), otherwise. Features selection (FS) was performed by a genetic algorithm and classification by a Support Vector Machine (SVM) classifier. Sensitivity, specificity, negative (NPV) and positive (PPV) predictive values were evaluated for all lesions in the training and validation sets, separately. On the training set, we obtained sensitivity of 86%, specificity of 67%, PPV of 89% and NPV of 61%, while, on the validation set, we reached a sensitivity of 73%, specificity of 47%, PPV of 64% and NPV of 57%. Specificity was biased by the low number of R- lesions on the validation set. The promising results obtained in the validation dataset should be extended to a larger cohort of patient to further validate our method.Clinical Relevance- to personalize treatment of patients with metastastic colorectal cancer, based on the likelihood of response to chemotherapy of each liver metastasis.

A Convolutional Neural Network based system for Colorectal cancer segmentation on MRI images.

The aim of the study is to present a new Convolutional Neural Network (CNN) based system for the automatic segmentation of the colorectal cancer. The algorithm implemented consists of several steps: a pre-processing to normalize and highlights the tumoral area, the classification based on CNNs, and a post-processing aimed at reducing false positive elements. The classification is performed using three CNNs: each of them classifies the same regions of interest acquired from three different MR sequences. The final segmentation mask is obtained by a majority voting. Performances were evaluated using a semi-automatic segmentation revised by an experienced radiologist as reference standard. The system obtained Dice Similarity Coefficient (DSC) of 0.60, Precision (Pr) of 0.76 and Recall (Re) of 0.55 on the testing set. After applying the leave-one-out validation, we obtained a median DSC=0.58, Pr=0.74, Re=0.54. The promising results obtained by this system, if validated on a larger dataset, could strongly improve personalized medicine.