Machine Learning and Radiomics Analysis for Tumor Budding Prediction in Colorectal Liver Metastases Magnetic Resonance Imaging Assessment.

PURPOSE: We aimed to assess the efficacy of machine learning and radiomics analysis using magnetic resonance imaging (MRI) with a hepatospecific contrast agent, in a pre-surgical setting, to predict tumor budding in liver metastases. METHODS: Patients with MRI in a pre-surgical setting were retrospectively enrolled. Manual segmentation was made by means 3D Slicer image computing, and 851 radiomics features were extracted as median values using the PyRadiomics Python package. Balancing was performed and inter- and intraclass correlation coefficients were calculated to assess the between observer and within observer reproducibility of all radiomics extracted features. A Wilcoxon-Mann-Whitney nonparametric test and receiver operating characteristics (ROC) analysis were carried out. Balancing and feature selection procedures were performed. Linear and non-logistic regression models (LRM and NLRM) and different machine learning-based classifiers including decision tree (DT), k-nearest neighbor (KNN) and support vector machine (SVM) were considered. RESULTS: The internal training set included 49 patients and 119 liver metastases. The validation cohort consisted of a total of 28 single lesion patients. The best single predictor to classify tumor budding was original_glcm_Idn obtained in the T1-W VIBE sequence arterial phase with an accuracy of 84%; wavelet_LLH_firstorder_10Percentile was obtained in the T1-W VIBE sequence portal phase with an accuracy of 92%; wavelet_HHL_glcm_MaximumProbability was obtained in the T1-W VIBE sequence hepatobiliary excretion phase with an accuracy of 88%; and wavelet_LLH_glcm_Imc1 was obtained in T2-W SPACE sequences with an accuracy of 88%. Considering the linear regression analysis, a statistically significant increase in accuracy to 96% was obtained using a linear weighted combination of 13 radiomic features extracted from the T1-W VIBE sequence arterial phase. Moreover, the best classifier was a KNN trained with the 13 radiomic features extracted from the arterial phase of the T1-W VIBE sequence, obtaining an accuracy of 95% and an AUC of 0.96. The validation set reached an accuracy of 94%, a sensitivity of 86% and a specificity of 95%. CONCLUSIONS: Machine learning and radiomics analysis are promising tools in predicting tumor budding. Considering the linear regression analysis, there was a statistically significant increase in accuracy to 96% using a weighted linear combination of 13 radiomics features extracted from the arterial phase compared to a single radiomics feature.

Radiomics and machine learning analysis by computed tomography and magnetic resonance imaging in colorectal liver metastases prognostic assessment.

OBJECTIVE: The aim of this study was the evaluation radiomics analysis efficacy performed using computed tomography (CT) and magnetic resonance imaging in the prediction of colorectal liver metastases patterns linked to patient prognosis: tumor growth front; grade; tumor budding; mucinous type. Moreover, the prediction of liver recurrence was also evaluated. METHODS: The retrospective study included an internal and validation dataset; the first was composed by 119 liver metastases from 49 patients while the second consisted to 28 patients with single lesion. Radiomic features were extracted using PyRadiomics. Univariate and multivariate approaches including machine learning algorithms were employed. RESULTS: The best predictor to identify tumor growth was the Wavelet_HLH_glcm_MaximumProbability with an accuracy of 84% and to detect recurrence the best predictor was wavelet_HLH_ngtdm_Complexity with an accuracy of 90%, both extracted by T1-weigthed arterial phase sequence. The best predictor to detect tumor budding was the wavelet_LLH_glcm_Imc1 with an accuracy of 88% and to identify mucinous type was wavelet_LLH_glcm_JointEntropy with an accuracy of 92%, both calculated on T2-weigthed sequence. An increase statistically significant of accuracy (90%) was obtained using a linear weighted combination of 15 predictors extracted by T2-weigthed images to detect tumor front growth. An increase statistically significant of accuracy at 93% was obtained using a linear weighted combination of 11 predictors by the T1-weigthed arterial phase sequence to classify tumor budding. An increase statistically significant of accuracy at 97% was obtained using a linear weighted combination of 16 predictors extracted on CT to detect recurrence. An increase statistically significant of accuracy was obtained in the tumor budding identification considering a K-nearest neighbors and the 11 significant features extracted T1-weigthed arterial phase sequence. CONCLUSIONS: The results confirmed the Radiomics capacity to recognize clinical and histopathological prognostic features that should influence the choice of treatments in colorectal liver metastases patients to obtain a more personalized therapy.

Multi-Institutional Evaluation of Pathologists' Assessment Compared to Immunoscore.

BACKGROUND: The Immunoscore (IS) is a quantitative digital pathology assay that evaluates the immune response in cancer patients. This study reports on the reproducibility of pathologists' visual assessment of CD3+- and CD8+-stained colon tumors, compared to IS quantification. METHODS: An international group of expert pathologists evaluated 540 images from 270 randomly selected colon cancer (CC) cases. Concordance between pathologists' T-score, corresponding hematoxylin-eosin (H&E) slides, and the digital IS was evaluated for two- and three-category IS. RESULTS: Non-concordant T-scores were reported in more than 92% of cases. Disagreement between semi-quantitative visual assessment of T-score and the reference IS was observed in 91% and 96% of cases before and after training, respectively. Statistical analyses showed that the concordance index between pathologists and the digital IS was weak in two- and three-category IS, respectively. After training, 42% of cases had a change in T-score, but no improvement was observed with a Kappa of 0.465 and 0.374. For the 20% of patients around the cut points, no concordance was observed between pathologists and digital pathology analysis in both two- and three-category IS, before or after training (all Kappa < 0.12). CONCLUSIONS: The standardized IS assay outperformed expert pathologists' T-score evaluation in the clinical setting. This study demonstrates that digital pathology, in particular digital IS, represents a novel generation of immune pathology tools for reproducible and quantitative assessment of tumor-infiltrated immune cell subtypes.

Radiomics and machine learning analysis based on magnetic resonance imaging in the assessment of liver mucinous colorectal metastases.

PURPOSE: The purpose of this study is to evaluate the Radiomics and Machine Learning Analysis based on MRI in the assessment of Liver Mucinous Colorectal Metastases.Query METHODS: The cohort of patients included a training set (121 cases) and an external validation set (30 cases) with colorectal liver metastases with pathological proof and MRI study enrolled in this approved study retrospectively. About 851 radiomics features were extracted as median values by means of the PyRadiomics tool on volume on interest segmented manually by two expert radiologists. Univariate analysis, linear regression modelling and pattern recognition methods were used as statistical and classification procedures. RESULTS: The best results at univariate analysis were reached by the wavelet_LLH_glcm_JointEntropy extracted by T2W SPACE sequence with accuracy of 92%. Linear regression model increased the performance obtained respect to the univariate analysis. The best results were obtained by a linear regression model of 15 significant features extracted by the T2W SPACE sequence with accuracy of 94%, a sensitivity of 92% and a specificity of 95%. The best classifier among the tested pattern recognition approaches was k-nearest neighbours (KNN); however, KNN achieved lower precision than the best linear regression model. CONCLUSIONS: Radiomics metrics allow the mucinous subtype lesion characterization, in order to obtain a more personalized approach. We demonstrated that the best performance was obtained by T2-W extracted textural metrics.

Radiomics and Machine Learning Analysis Based on Magnetic Resonance Imaging in the Assessment of Colorectal Liver Metastases Growth Pattern.

To assess Radiomics and Machine Learning Analysis in Liver Colon and Rectal Cancer Metastases (CRLM) Growth Pattern, we evaluated, retrospectively, a training set of 51 patients with 121 liver metastases and an external validation set of 30 patients with a single lesion. All patients were subjected to MRI studies in pre-surgical setting. For each segmented volume of interest (VOI), 851 radiomics features were extracted using PyRadiomics package. Nonparametric test, univariate, linear regression analysis and patter recognition approaches were performed. The best results to discriminate expansive versus infiltrative front of tumor growth with the highest accuracy and AUC at univariate analysis were obtained by the wavelet_LHH_glrlm_ShortRunLowGray Level Emphasis from portal phase of contrast study. With regard to linear regression model, this increased the performance obtained respect to the univariate analysis for each sequence except that for EOB-phase sequence. The best results were obtained by a linear regression model of 15 significant features extracted by the T2-W SPACE sequence. Furthermore, using pattern recognition approaches, the diagnostic performance to discriminate the expansive versus infiltrative front of tumor growth increased again and the best classifier was a weighted KNN trained with the 9 significant metrics extracted from the portal phase of contrast study, with an accuracy of 92% on training set and of 91% on validation set. In the present study, we have demonstrated as Radiomics and Machine Learning Analysis, based on EOB-MRI study, allow to identify several biomarkers that permit to recognise the different Growth Patterns in CRLM.

Validation of the standardized index of shape tool to analyze DCE-MRI data in the assessment of neo-adjuvant therapy in locally advanced rectal cancer.

PURPOSE: Standardized index of shape (SIS) tool validation to examine dynamic contrast enhanced-magnetic resonance imaging (DCE-MRI) in preoperative chemo-radiation therapy (pCRT) assessment of locally advanced rectal cancer (LARC) in order to guide the surgeon versus more or less conservative treatment. MATERIALS AND METHODS: A total of 194 patients (January 2008-November 2020), with III-IV locally advanced rectal cancer and subjected to pCRT were included. Three expert radiologists performed DCE-MRI analysis using SIS tool. Degree of absolute agreement among measurements, degree of consistency among measurements, degree of reliability and level of variability were calculated. Patients with a pathological tumour regression grade (TRG) 1 or 2 were classified as major responders (complete responders have TRG 1). RESULTS: Good significant correlation was obtained between SIS measurements (range 0.97-0.99). The degree of absolute agreement ranges from 0.93 to 0.99, the degree of consistency from 0.81 to 0.9 and the reliability from 0.98 to 1.00 (p value < < 0.001). The variability coefficient ranges from 3.5% to 26%. SIS value obtained to discriminate responders by non-responders a sensitivity of 95.9%, a specificity of 84.7% and an accuracy of 91.8% while to detect complete responders, a sensitivity of 99.2%, a specificity of 63.9% and an accuracy of 86.1%. CONCLUSION: SIS tool is suitable to assess pCRT response both to identify major responders and complete responders in order to guide the surgeon versus more or less conservative treatment.

Magnetic resonance imaging in the assessment of pancreatic cancer with quantitative parameter extraction by means of dynamic contrast-enhanced magnetic resonance imaging, diffusion kurtosis imaging and intravoxel incoherent motion diffusion-weighted imaging.

BACKGROUND: Despite great technical advances in imaging, such as multidetector computed tomography and magnetic resonance imaging (MRI), diagnosing pancreatic solid lesions correctly remains challenging, due to overlapping imaging features with benign lesions. We wanted to evaluate functional MRI to differentiate pancreatic tumors, peritumoral inflammatory tissue, and normal pancreatic parenchyma by means of dynamic contrast-enhanced MRI (DCE-MRI)-, diffusion kurtosis imaging (DKI)-, and intravoxel incoherent motion model (IVIM) diffusion-weighted imaging (DWI)-derived parameters. METHODS: We retrospectively analyzed 24 patients, each with histopathological diagnosis of pancreatic tumor, and 24 patients without pancreatic lesions. Functional MRI was acquired using a 1.5 MR scanner. Peritumoral inflammatory tissue was assessed by drawing regions of interest on the tumor contours. DCE-MRI, IVIM and DKI parameters were extracted. Nonparametric tests and receiver operating characteristic (ROC) curves were calculated. RESULTS: There were statistically significant differences in median values among the three groups observed by Kruskal-Wallis test for the DKI mean diffusivity (MD), IVIM perfusion fraction (fp) and IVIM tissue pure diffusivity (Dt). MD had the best results to discriminate normal pancreas plus peritumoral inflammatory tissue versus pancreatic tumor, to separate normal pancreatic parenchyma versus pancreatic tumor and to differentiate peritumoral inflammatory tissue versus pancreatic tumor, respectively, with an accuracy of 84%, 78%, 83% and area under ROC curve (AUC) of 0.85, 0.82, 0.89. The findings were statistically significant compared with those of other parameters (p value < 0.05 using McNemar's test). Instead, to discriminate normal pancreas versus peritumoral inflammatory tissue or pancreatic tumor and to differentiate normal pancreatic parenchyma versus peritumoral inflammatory tissue, there were no statistically significant differences between parameters' accuracy (p > 0.05 at McNemar's test). CONCLUSIONS: Diffusion parameters, mainly MD by DKI, could be helpful for the differentiation of normal pancreatic parenchyma, perilesional inflammation, and pancreatic tumor.

Major and ancillary features according to LI-RADS in the assessment of combined hepatocellular-cholangiocarcinoma.

Background The aim of the study was to investigate the performance of the Liver Imaging Reporting and Data System (LI-RADS) v2018 for combined hepatocellular-cholangiocarcinoma (cHCC-CCA) identifying the features that allow an accurate characterization. Patients and methods Sixty-two patients (median age, 63 years; range, 38-80 years), with pre-surgical biopsy diagnosis of hepatocellular carcinoma (HCC) that underwent hepatic resection, comprised our retrospective study. All patients were subject to multidetector computed tomography (MDCT); 23 patients underwent to magnetic resonance (MR) study. The radiologist reported the presence of the HCC by using LIRADS v2018 assessing major and ancillary features. Results Final histological diagnosis was HCC for 51 patients and cHCC-CCA for 11 patients. The median nodule size was 46.0 mm (range 10-190 mm). For cHCC-CCA the median size was 33.5 mm (range 20-80 mm), for true HCC the median size was 47.5 mm (range 10-190 mm). According to LIRADS categories: 54 (87.1%) nodules as defined as LR-5, 1 (1.6%) as LR-3, and 7 (11.3%) as LR-M. Thirty-nine nodules (63%) showed hyper-enhancement in arterial phase; among them 4 were cHCC-CCA (36.4% of cHCC-CCA) and 35 (68.6%) true HCC. Forty-three nodules (69.3%) showed washout appearance; 6 cHCC-CCAs (54.5% of cHCC-CCA) and 37 true HCC (72.5%) had this feature. Only two cHCC-CCA patients (18.2% of cHCC-CCA) showed capsule appearance. Five cHCC-CCA (71.4% of cHCC-CCA) showed hyperintensity on T2-W sequences while two (28.6%) showed inhomogeneous signal in T2-W. All cHCC-CCA showed restricted diffusion. Seven cHCC-CCA patients showed a progressive contrast enhancement and satellite nodules. Conclusions The presence of satellite nodules, hyperintense signal on T2-W, restricted diffusion, the absence of capsule appearance in nodule that shows peripheral and progressive contrast enhancement are suggestive features of cHCC-CCA.

Diagnostic performance of gadoxetic acid-enhanced liver MRI versus multidetector CT in the assessment of colorectal liver metastases compared to hepatic resection.

BACKGROUND: Imaging is an essential tool in the management of patients with Colorectal cancer (CRC) by helping evaluate number and sites of metastases, determine resectability, assess response to treatment, detect drug toxicities and recurrences. Although multidetector computed tomography (MDCT) is the first tool used for staging and patient's surveillance, magnetic resonance imaging (MRI) is the most reliable imaging modality that allows to assess liver metastases. Our purpose is to compare the diagnostic performance of gadoxetic acid-(Gd-EOB) enhanced liver MRI and contrast-enhanced MDCT in the detection of liver metastasis from colorectal cancer (mCRC). METHODS: One hundred and twenty-eight patients with pathologically proven mCRC (512 liver metastases) underwent Gd-EOB MRI and MDCT imaging. An additional 46 patients without mCRC were included as control subjects. Three radiologists independently graded the presence of liver nodules on a five-point confidence scale. Sensitivity and specificity for the detection of metastases were calculated. Weighted к values were used to evaluate inter-reader agreement of the confidence scale regarding the presence of the lesion. RESULTS: MRI detected 489 liver metastases and MDCT 384. In terms of per-lesion sensitivity in the detection of liver metastasis, all three readers had higher diagnostic sensitivity with Gd-EOB MRI than with MDCT (95.5% vs. 72% reader 1; 90% vs. 72% reader 2; 96% vs. 75% reader 3). Each reader showed a statistical significant difference (p < <.001 at Chi square test). MR imaging showed a higher performance than MDCT in per-patient detection sensitivity (100% vs. 74.2% [p < <.001] reader 1, 98% vs. 73% [p < <.001] reader 2, and 100% vs. 78% [p < <.001] reader 3). In the control group, MRI and MDCT showed similar per-patient specificity (100% vs. 98% [p = 0.31] reader 1, 100% vs. 100% [p = 0.92] reader 2, and 100% vs. 96% [p = 0.047] reader 3). Inter-reader agreement of lesion detection between the three radiologists was moderate to excellent (k range, 0.56-0.86) for Gd-EOB MRI and substantial to excellent for MDCT (k range, 0.75-0.8). CONCLUSION: Gadoxetic acid-enhanced MRI performs significantly better in the detection of mCRC, than MDCT, particularly in patients treated with chemotherapy, in subcapsular lesions, and in peribiliary metastases.

Diagnostic accuracy of magnetic resonance, computed tomography and contrast enhanced ultrasound in radiological multimodality assessment of peribiliary liver metastases.

PURPOSE: We compared diagnostic performance of Magnetic Resonance (MR), Computed Tomography (CT) and Ultrasound (US) with (CEUS) and without contrast medium to identify peribiliary metastasis. METHODS: We identified 35 subjects with histological proven peribiliary metastases who underwent CEUS, CT and MR study. Four radiologists evaluated the presence of peribiliary lesions, using a 4-point confidence scale. Echogenicity, density and T1-Weigthed (T1-W), T2-W and Diffusion Weighted Imaging (DWI) signal intensity as well as the enhancement pattern during contrast studies on CEUS, CT and MR so as hepatobiliary-phase on MRI was assessed. RESULTS: All lesions were detected by MR. CT detected 8 lesions, while US/CEUS detected one lesion. According to the site of the lesion, respect to the bile duct and hepatic parenchyma: 19 (54.3%) were periductal, 15 (42.8%) were intra-periductal and 1 (2.8%) was periductal-intrahepatic. According to the confidence scale MRI had the best diagnostic performance to assess the lesion. CT obtained lower diagnostic performance. There was no significant difference in MR signal intensity and contrast enhancement among all metastases (p>0.05). There was no significant difference in CT density and contrast enhancement among all metastases (p>0.05). CONCLUSIONS: MRI is the method of choice for biliary tract tumors but it does not allow a correct differential diagnosis among different histological types of metastasis. The presence of biliary tree dilatation without hepatic lesions on CT and US/CEUS study may be an indirect sign of peribiliary metastases and for this reason the patient should be evaluated by MRI.

Early FDG PET response assessment of preoperative radiochemotherapy in locally advanced rectal cancer: correlation with long-term outcome.

PURPOSE: The aim of the present study is to prospectively evaluate the prognostic value of previously defined [(18)F]2-fluoro-2-deoxy-D-glucose positron emission tomography (FDG PET) criteria of early metabolic response in patients with locally advanced rectal cancer (LARC) after long-term follow-up. METHODS: Forty-two patients with poor prognosis LARC underwent three biweekly courses of chemotherapy with oxaliplatin, raltitrexed and 5-fluorouracil modulated by levofolinic acid during pelvic radiotherapy. FDG PET studies were performed before and 12 days after the beginning of the chemoradiotherapy (CRT) treatment. Total mesorectal excision (TME) was carried out 8 weeks after completion of CRT. A previously identified cutoff value of ≥52 % reduction of the baseline mean FDG standardized uptake value (SUV(mean)) was applied to differentiate metabolic responders from non-responders and correlated to tumour regression grade (TRG) and survival. RESULTS: Twenty-two metabolic responders showed complete (TRG1) or subtotal tumour regression (TRG2) and demonstrated a statistically significantly higher 5-year relapse-free survival (RFS) compared with the 20 non-responders (86 vs 55 %, p = .014) who showed TRG3 and TRG4 pathologic responses. A multivariate analysis demonstrated that early ∆SUV(mean) was the only pre-surgical parameter correlated to the likelihood of recurrence (p = .05). CONCLUSION: This study is the first prospective long-term evaluation demonstrating that FDG PET is not only an early predictor of pathologic response but is also a valuable prognostic tool. Our results indicate the potential of FDG PET for optimizing multidisciplinary management of patients with LARC.