Prediction of tumor grade and lymphovascular space invasion in endometrial adenocarcinoma with MR imaging-based radiomic analysis

PurposeTo evaluate the capabilities of two-dimensional magnetic resonance imaging (MRI)-based texture analysis features, tumor volume, tumor short axis and apparent diffusion coefficient (ADC) in predicting histopathological high-grade and lymphovascular space invasion (LVSI) in endometrial adenocarcinoma.Materials and methodsSeventy-three women (mean age: 66 ± 11.5 [SD] years; range: 45–88 years) with endometrial adenocarcinoma who underwent MRI of the pelvis at 1.5-T before hysterectomy were retrospectively included. Texture analysis was performed using TexRAD® software on T2-weighted images and ADC maps. Primary outcomes were high-grade and LVSI prediction using histopathological analysis as standard of reference. After data reduction using ascending hierarchical classification analysis, a predictive model was obtained by stepwise multivariate logistic regression and performances were assessed using cross-validated receiver operator curve (ROC).ResultsA total of 72 texture features per tumor were computed. Texture model yielded 52% sensitivity and 75% specificity for the diagnosis of high-grade tumor (areas under ROC curve [AUC] = 0.64) and 71% sensitivity and 59% specificity for the diagnosis of LVSI (AUC = 0.59). Volumes and tumor short axis were greater for high-grade tumors (P = 0.0002 and P = 0.004, respectively) and for patients with LVSI (P = 0.004 and P = 0.0279, respectively). No differences in ADC values were found between high-grade and low-grade tumors and for LVSI. A tumor short axis ≥ 20 mm yielded 95% sensitivity and 75% specificity for the diagnosis of high-grade tumor (AUC = 0.86).ConclusionMRI-based texture analysis is of limited value to predict high grade and LVSI of endometrial adenocarcinoma. A tumor short axis ≥ 20 mm is the best predictor of high grade and LVSI.     

Differentiating autoimmune pancreatitis from pancreatic ductal adenocarcinoma with CT radiomics features

PurposeThe purpose of this study was to determine whether computed tomography (CT)-based machine learning of radiomics features could help distinguish autoimmune pancreatitis (AIP) from pancreatic ductal adenocarcinoma (PDAC).Materials and MethodsEighty-nine patients with AIP (65 men, 24 women; mean age, 59.7 ± 13.9 [SD] years; range: 21–83 years) and 93 patients with PDAC (68 men, 25 women; mean age, 60.1 ± 12.3 [SD] years; range: 36–86 years) were retrospectively included. All patients had dedicated dual-phase pancreatic protocol CT between 2004 and 2018. Thin-slice images (0.75/0.5 mm thickness/increment) were compared with thick-slices images (3 or 5 mm thickness/increment). Pancreatic regions involved by PDAC or AIP (areas of enlargement, altered enhancement, effacement of pancreatic duct) as well as uninvolved parenchyma were segmented as three-dimensional volumes. Four hundred and thirty-one radiomics features were extracted and a random forest was used to distinguish AIP from PDAC. CT data of 60 AIP and 60 PDAC patients were used for training and those of 29 AIP and 33 PDAC independent patients were used for testing.ResultsThe pancreas was diffusely involved in 37 (37/89; 41.6%) patients with AIP and not diffusely in 52 (52/89; 58.4%) patients. Using machine learning, 95.2% (59/62; 95% confidence interval [CI]: 89.8–100%), 83.9% (52:67; 95% CI: 74.7–93.0%) and 77.4% (48/62; 95% CI: 67.0–87.8%) of the 62 test patients were correctly classified as either having PDAC or AIP with thin-slice venous phase, thin-slice arterial phase, and thick-slice venous phase CT, respectively. Three of the 29 patients with AIP (3/29; 10.3%) were incorrectly classified as having PDAC but all 33 patients with PDAC (33/33; 100%) were correctly classified with thin-slice venous phase with 89.7% sensitivity (26/29; 95% CI: 78.6–100%) and 100% specificity (33/33; 95% CI: 93–100%) for the diagnosis of AIP, 95.2% accuracy (59/62; 95% CI: 89.8–100%) and area under the curve of 0.975 (95% CI: 0.936–1.0).ConclusionsRadiomic features help differentiate AIP from PDAC with an overall accuracy of 95.2%.     

CT features of lung metastases from pancreatic adenocarcinoma: Correlation with histopathologic findings

PurposeThe purpose of this study was to evaluate the prevalence of an atypical, alveolar presentation of pulmonary metastases from pancreatic adenocarcinoma (PDAC) on computed tomography (CT) and to correlate CT features with those obtained at histopathologic analysis.Material and methodsA total of 76 patients with lung metastases from PDAC over a 10-year period (2009–2019) in a French university hospital were retrospectively included. There were 34 men and 42 women with a mean age of 67.6 ± 11.3 (SD) years (range: 38–89 years). CT features of PDAC were classified according to their presentations as usual metastatic pattern or atypical alveolar pattern; the atypical alveolar pattern corresponding to either ground glass nodules or opacities, solid nodules with a halo sign, “air-space” nodules with air bronchogram, or parenchymal consolidation. Imaging-histopathologic correlation was performed when tissue samples were available.ResultsPulmonary metastases were synchronous in 36 patients (36/76; 47%) and metachronous in 40 patients (40/76; 53%). A predominant alveolar presentation on CT was observed in 17 patients (17/76, 22%). Nodules with halo sign were the predominant alveolar pattern in 7 patients (7/17; 41%), air-space nodules were predominant in 4 patients (4/17; 24%) whereas pure ground glass nodules and consolidations were observed as predominant features in 3 patients (3/17; 18%) each. For 5 patients who had histopathological confirmation, alveolar metastases of PDAC were characterized by columnar tumor cells lining the alveolar wall, which was not seen in other radiological presentations, whereas there were no differences regarding mucin secretion between pulmonary metastases with alveolar presentation and those with typical pattern.ConclusionsLung metastases from PDAC may present with a so-called “alveolar” pattern on CT. This misleading CT features is found in 22% of patients with lung metastases from PDAC and is due to lepidic growth of the metastatic cells.     

3D reconstructions, 4D imaging and postprocessing with CT in musculoskeletal disorders: Past,present and future

Three-dimensional (3D) imaging and post processing are common tasks used daily in many disciplines. The purpose of this article is to review the new postprocessing tools available. Although 3D imaging can be applied to all anatomical regions and used with all imaging techniques, its most varied and relevant applications are found with computed tomography (CT) data in musculoskeletal imaging. These new applications include global illumination rendering (GIR), unfolded rib reformations, subtracted CT angiography for bone analysis, dynamic studies, temporal subtraction and image fusion. In all of these tasks, registration and segmentation are two basic processes that affect the quality of the results. GIR simulates the complete interaction of photons with the scanned object, providing photorealistic volume rendering. Reformations to unfold the rib cage allow more accurate and faster diagnosis of rib lesions. Dynamic CT can be applied to cinematic joint evaluations a well as to perfusion and angiographic studies. Finally, more traditional techniques, such as minimum intensity projection, might find new applications for bone evaluation with the advent of ultra-high-resolution CT scanners. These tools can be used synergistically to provide morphologic, topographic and functional information and increase the versatility of CT.     

Comparison of conventional magnetic resonance imaging and diffusion-weighted imaging in the differentiation of bone plasmacytoma from bone metastasis in the extremities

PurposeTo compare conventional magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI) in the differentiation of bone plasmacytoma from bone metastasis in the extremities.Materials and methodsA total of 65 patients with 27 bone plasmacytomas (11 men; mean age, 63.6 ± 8.2 [SD] years) and 38 patients with bone metastases (20 men; mean age, 64.1 ± 11.5 [SD] years) were retrospectively included. Plasmacytomas and metastases were compared for size, peritumoral edema, signal intensity (SI), SI pattern, apparent diffusion coefficient (ADC) values and standard deviation (SD) of ADC. Receiver operating characteristic analysis with area under the curve (AUC) was used to calculate sensitivity, specificity, and accuracy of MRI and DWI for the diagnosis of plasmacytoma according to a defined cut-off value.ResultsOn conventional MRI, plasmacytomas showed less peritumoral edema (22% vs. 71%; P < 0.001), were more often hyperintense on T1-weighted image (48% vs. 18%; P = 0.022) and more homogeneous on T2-weighted image (78% vs. 26%; P < 0.001) and contrast-enhanced T1-weighted images (70% vs. 25%; P = 0.001) than bone metastases. Mean ADC value and SD of ADC were significantly lower in bone plasmacytomas (760.1 ± 196.9 [SD] μm²/s and 161.5 ± 62.7 [SD], respectively) than in bone metastases (1214.2 ± 382.6 [SD] μm²/s and 277.0 ± 110.3 [SD], respectively) (P < 0.001). Using an ADC value ≤ 908.3 μm²/s, DWI yielded 88% sensitivity and 78% specificity for the diagnosis of plasmacytoma. ADC value yielded best area under the curve (AUC = 0.913), followed by SD of ADC (AUC = 0.814) and homogeneity on T2-weighted images (AUC = 0.757). The combination of conventional MRI and DWI (AUC = 0.894) showed improved diagnostic performance over conventional MRI alone (AUC= 0.843) for discriminating between plasmacytoma and metastasis.ConclusionConventional MRI in combination with DWI can be useful to discriminate between bone plasmacytoma and bone metastasis in the extremities.     

Automatic coronary artery calcium scoring from unenhanced-ECG-gated CT using deep learning

PurposeThe purpose of this study was to develop and evaluate an algorithm that can automatically estimate the amount of coronary artery calcium (CAC) from unenhanced electrocardiography (ECG)-gated computed tomography (CT) cardiac volume acquisitions by using convolutional neural networks (CNN).Materials and methodsThe method used a set of five CNN with three-dimensional (3D) U-Net architecture trained on a database of 783 CT examinations to detect and segment coronary artery calcifications in a 3D volume. The Agatston score, the conventional CAC scoring, was then computed slice by slice from the resulting segmentation mask and compared to the ground truth manually estimated by radiologists. The quality of the estimation was assessed with the concordance index (C-index) on CAC risk category on a separate testing set of 98 independent CT examinations.ResultsThe final model yielded a C-index of 0.951 on the testing set. The remaining errors of the method were mainly observed on small-size and/or low-density calcifications, or calcifications located near the mitral valve or ring.ConclusionThe deep learning-based method proposed here to compute automatically the CAC score from unenhanced-ECG-gated cardiac CT is fast, robust and yields accuracy similar to those of other artificial intelligence methods, which could improve workflow efficiency, eliminating the time spent on manually selecting coronary calcifications to compute the Agatston score.     

Automatic cervical lymphadenopathy segmentation from CT data using deep learning

PurposeThe purpose of this study was to develop a fast and automatic algorithm to detect and segment lymphadenopathy from head and neck computed tomography (CT) examination.Materials and methodsAn ensemble of three convolutional neural networks (CNNs) based on a U-Net architecture were trained to segment the lymphadenopathies in a fully supervised framework. The resulting predictions were assessed using the Dice similarity coefficient (DSC) on examinations presenting one or more adenopathies. On examinations without adenopathies, the score was given by the formula M/(M + A) where M was the mean adenopathy volume per patient and A the volume segmented by the algorithm. The networks were trained on 117 annotated CT acquisitions.ResultsThe test set included 150 additional CT acquisitions unseen during the training. The performance on the test set yielded a mean score of 0.63.ConclusionDespite limited available data and partial annotations, our CNN based approach achieved promising results in the task of cervical lymphadenopathy segmentation. It has the potential to bring precise quantification to the clinical workflow and to assist the clinician in the detection task.     

CT evaluation of patent artery after percutaneous cryoablation of renal cell carcinoma

PurposeThe purpose of this retrospective study was to determine the incidence of persistent patent artery after percutaneous cryoablation of renal cell carcinoma (RCC) and the relationship between patent arteries one month after cryoablation and early tumor progression.Materials and methodsOne hundred and fifty-nine patients (112 men, 47 women; mean age, 63.6 ± 14.6 [SD] years; age range: 21–91 years) who underwent percutaneous cryoablation for 186 RCCs (mean diameter, 1.9 ± 0.6 [SD] cm; range: 0.7–4.0 cm) were retrospectively included. After cryoablation, patients underwent contrast-enhanced computed tomography (CT) with ≤ 2-mm slice thickness within one week from cryoablation, and at one, three, and six months. The time course of patent artery in the ablated renal parenchyma after cryoablation was the primary endpoint. The relationships between patent arteries one month after cryoablation and treatment effectiveness, tumor vascularity, tumor enhancement one month after cryoablation, tumor subtype, and renal function changes were evaluated as secondary endpoints.ResultsCT showed patent arteries in the ablated renal parenchyma within one week in 166 RCCs (89.2%), at one month in 54 RCCs (29.0%), at three months in 8 RCCs (4.3%), and at six months in 2 RCCs (1.1%). The presence of patent artery one month after cryoablation was significantly associated with tumor enhancement at the same time point (P = 0.015). There was no association between patent arteries one month after cryoablation and treatment effectiveness (P = 0.693).ConclusionPatent arteries in the ablated renal parenchyma are commonly observed on CT examination after percutaneous cryoablation of RCC. However, they gradually disappear and do not require specific treatment.     

Prediction of overall survival in patients with hepatocellular carcinoma treated with Y-90 radioembolization by imaging response criteria

PurposeTo evaluate the potential of imaging criteria in predicting overall survival of patients with hepatocellular carcinoma (HCC) after a first transcatheter arterial yttrium-90 radioembolization (TARE)Materials and methodsFrom October 2013 to July 2017, 37 patients with HCC were retrospectively included. There were 34 men and 3 women with a mean age of 60.5 ± 10.2 (SD) years (range: 32.7–78.9 years). Twenty-five patients (68%) were Barcelona Clinic Liver Cancer (BCLC) C and 12 (32%) were BCLC B. Twenty-four primary index tumors (65%) were > 5 cm. Three radiologists evaluated tumor response on pre- and 4–7 months post-TARE magnetic resonance imaging or computed tomography examinations, using Response Evaluation Criteria in Solid Tumors (RECIST) 1.1, modified RECIST (mRECIST), European Association for Study of the Liver (EASL), volumetric RECIST (vRECIST), quantitative EASL (qEASL) and the Liver Imaging Reporting and Data System treatment response algorithm. Kaplan–Meier survival curves were used to compare responders and non-responders for each criterion. Univariate and multivariate Cox proportional hazard ratio (HR) analysis were used to identify covariates associated with overall survival. Fleiss kappa test was used to assess interobserver agreement.ResultsAt multivariate analysis, RECIST 1.1 (HR: 0.26; 95% confidence interval [95% CI]: 0.09–0.75; P = 0.01), mRECIST (HR: 0.22; 95% CI: 0.08–0.59; P = 0.003), EASL (HR: 0.22; 95% CI: 0.07–0.63; P = 0.005), and qEASL (HR: 0.30; 95% CI: 0.12–0.80; P = 0.02) showed a significant difference in overall survival between responders and nonresponders. RECIST 1.1 had the highest interobserver reproducibility.ConclusionRECIST and mRECIST seem to be the best compromise between reproducibility and ability to predict overall survival in patients with HCC treated with TARE.     

Feasibility of lung imaging with a large field-of-view spectral photon-counting CT system

PurposeThe purpose of this study was to characterize the technical capabilities and feasibility of a large field-of-view clinical spectral photon-counting computed tomography (SPCCT) prototype for high-resolution (HR) lung imaging.Materials and methodsMeasurement of modulation transfer function (MTF) and acquisition of a line pairs phantom were performed. An anthropomorphic lung nodule phantom was scanned with standard (120 kVp, 62 mAs), low (120 kVp, 11 mAs), and ultra-low (80 kVp, 3 mAs) radiation doses. A human volunteer underwent standard (120 kVp, 63 mAs) and low (120 kVp, 11 mAs) dose scans after approval by the ethics committee. HR images were reconstructed with 1024 matrix, 300 mm field of view and 0.25 mm slice thickness using a filtered-back projection (FBP) and two levels of iterative reconstruction (iDose 5 and 9). The conspicuity and sharpness of various lung structures (distal airways, vessels, fissures and proximal bronchial wall), image noise, and overall image quality were independently analyzed by three radiologists and compared to a previous HR lung CT examination of the same volunteer performed with a conventional CT equipped with energy integrating detectors (120 kVp, 10 mAs, FBP).ResultsTen percent MTF was measured at 22.3 lp/cm with a cut-off at 31 lp/cm. Up to 28 lp/cm were depicted. While mixed and solid nodules were easily depicted on standard and low-dose phantom images, higher iDose levels and slice thicknesses (1 mm) were needed to visualize ground-glass components on ultra-low-dose images. Standard dose SPCCT images of in vivo lung structures were of greater conspicuity and sharpness, with greater overall image quality, and similar image noise (despite a flux reduction of 23%) to conventional CT images. Low-dose SPCCT images were of greater or similar conspicuity and sharpness, similar overall image quality, and lower but acceptable image noise (despite a flux reduction of 89%).ConclusionsA large field-of-view SPCCT prototype demonstrates HR technical capabilities and high image quality for high resolution lung CT in human.     

Single-source dual energy CT to assess myocardial extracellular volume fraction in aortic stenosis before transcatheter aortic valve implantation (TAVI)

PurposeTo assess myocardial extracellular volume fraction (ECV) measurement provided by a single-source dual-energy computed tomography (SSDE-CT) acquisition added at the end of a routine CT examination before transcatether aortic valve implantation (TAVI) compared to cardiac magnetic resonance imaging (MRI).Materials and methodsTwenty-one patients (10 men, 11 women; mean age, 86 ± 4.9 years [SD]; age range: 71–92 years) with severe aortic stenosis underwent standard pre-TAVI CT with additional cardiac SSDE-CT acquisition 7 minutes after intravenous administration of iodinated contrast material and myocardial MRI including pre- and post-contrast T1-maps. Myocardial ECV and standard deviation (σECV) were calculated in the 16-segments model. ECV provided by SSDE-CT was compared to ECV provided by MRI, which served as the reference. Analyses were performed on a per-segment basis and on a per-patient involving the mean value of the 16-segments.ResultsECV was slightly overestimated by SSDE-CT (29.9 ± 4.6 [SD] %; range: 20.9%–48.3%) compared to MRI (29.1 ± 3.9 [SD] %; range: 22.0%–50.7%) (P < 0.0001) with a bias and limits of agreement of +2.3% (95%CI: −16.1%– + 20.6%) and +2.5% (95%CI: −2.1%– + 7.1%) for per-segment and per-patient-analyses, respectively. Good (r = 0.81 for per-segment-analysis) to excellent (r = 0.97 for per-patient-analysis) linear relationships (both P < 0.0001) were obtained. The σECV was significantly higher at SSDE-CT (P < 0.0001). Additional radiation dose from CT was 1.89 ± 0.38 (SD) mSv (range: 1.48–2.47 mSv).ConclusionA single additional SSDE-CT acquisition added at the end of a standard pre-TAVI CT protocol can provide ECV measurement with good to excellent linear relationship with MRI.     

Acute intracranial hypertension management in metastatic brain tumor: A French national survey

BackgroundBrain metastases occur in 15–30% of cancer patients and their frequency has increased over time. They can cause intracranial hypertension, even in the absence of hydrocephalus. Emergency surgical management of brain metastasis-related intracranial hypertension is not guided by specific recommendations.ObjectiveWe aimed to make a French national survey of emergency management of intracranial hypertension without hydrocephalus in the context of cerebral metastasis.MethodsA national online survey of French neurosurgeons from 16 centers was conducted, consisting of three clinical files, with multiple-choice questions on diagnostic and therapeutic management in different emergency situations.ResultsIn young patients without any previously known primary cancer, acute intracranial hypertension due to a seemingly metastatic single brain tumor indicated emergency surgery for all those interviewed; 61% aimed at complete resection; brain MRI was mandatory for 74%. When a primary cancer was known, 74% of respondents were more likely to propose surgery if an oncologist confirmed the possibility of adjuvant treatment; 27% were more likely to operate on an emergency basis when resection was scheduled after multi-disciplinary discussion, prior to acute degradation.ConclusionCurrently, there is no consensus on the emergency management of intracranial hypertension in metastatic brain tumor patients. In case of previously known primary cancer, a discussion with the oncology team seems necessary, even in emergency. Decision criteria emerge from our literature review, but require analysis in further studies.     

Effect of tin filter-based spectral shaping CT on image quality and radiation dose for routine use on ultralow-dose CT protocols: A phantom study

PurposeThe purpose of this study was to assess the impact of tin filter (TF) on X-ray beam quality, image quality and radiation dose and its suitability for routine use for chest and lumbar-spine/pelvis-hip ultralow-dose (ULD) CT examination protocols.Materials and methodsThe X-ray beam quality was determined by measuring the half-value layer (HVL) and calculating the mean weighted energy for 100, 120, 150 kVp (using standard filtration) and for 100 and 150 kVp using TF (Sn100 kVp and Sn150 kVp, respectively). Acquisitions were performed on a phantom at four dose levels for each previously defined kVp. The mean attenuation (N_CT), noise-power-spectrum (NPS) and task-based transfer function (TTF) were computed. The detectability index (d’) was computed to model the detection of two lesions in spine and pelvic/hip examination and two for chest exploration. Image quality and detectability using a TF were assessed for two routinely used ULD protocols.ResultsThe HVL and mean weighted energy increased using a TF for the same tube voltage. Using a TF for the same tube voltage changed N_CT for bone and acrylic inserts, decreased the NPS peak without changing the NPS spatial frequency and increased the TTF values. The d’ values were improved using a TF and with the dose increase. d’ values of all modeled lesions were improved using Sn100 kVp and Sn150 kVp for the lumbar-spine/pelvis-hip and chest ULD protocols except for sclerotic bone lesion using Sn150 kVp.ConclusionThe use of TF increases the X-ray beam quality and improves the image quality characteristics in phantom images, thus appearing as a promising tool for reducing dose and/or improving the image quality of ULD protocols.     

Abdominal musculature segmentation and surface prediction from CT using deep learning for sarcopenia assessment

PurposeThe purpose of this study was to build and train a deep convolutional neural networks (CNN) algorithm to segment muscular body mass (MBM) to predict muscular surface from a two-dimensional axial computed tomography (CT) slice through L3 vertebra.Materials and methodsAn ensemble of 15 deep learning models with a two-dimensional U-net architecture with a 4-level depth and 18 initial filters were trained to segment MBM. The muscular surface values were computed from the predicted masks and corrected with the algorithm's estimated bias. Resulting mask prediction and surface prediction were assessed using Dice similarity coefficient (DSC) and root mean squared error (RMSE) scores respectively using ground truth masks as standards of reference.ResultsA total of 1025 individual CT slices were used for training and validation and 500 additional axial CT slices were used for testing. The obtained mean DSC and RMSE on the test set were 0.97 and 3.7 cm² respectively.ConclusionDeep learning methods using convolutional neural networks algorithm enable a robust and automated extraction of CT derived MBM for sarcopenia assessment, which could be implemented in a clinical workflow.     

Artificial intelligence solution to classify pulmonary nodules on CT

PurposeThe purpose of this study was to create an algorithm to detect and classify pulmonary nodules in two categories based on their volume greater than 100 mm³ or not, using machine learning and deep learning techniques.Materials and methodThe dataset used to train the model was provided by the organization team of the SFR (French Radiological Society) Data Challenge 2019. An asynchronous and parallel 3-stages pipeline was developed to process all the data (a data “pre-processing” stage; a “nodule detection” stage; a “classifier” stage). Lung segmentation was achieved using 3D U-NET algorithm; nodule detection was done using 3D Retina-UNET and classifier stage with a support vector machine algorithm on selected features. Performances were assessed using area under receiver operating characteristics curve (AUROC).ResultsThe pipeline showed good performance for pathological nodule detection and patient diagnosis. With the preparation dataset, an AUROC of 0.9058 (95% confidence interval [CI]: 0.8746–0.9362) was obtained, 87% yielding accuracy (95% CI: 84.83%–91.03%) for the “nodule detection” stage, corresponding to 86% specificity (95% CI: 82%–92%) and 89% sensitivity (95% CI: 84.83%–91.03%).ConclusionA fully functional pipeline using 3D U-NET, 3D Retina-UNET and classifier stage with a support vector machine algorithm was developed, resulting in high capabilities for pulmonary nodule classification.     

MRI-based whole-tumor radiomics to classify the types of pediatric posterior fossa brain tumor

BackgroundDifferential diagnosis between medulloblastoma (MB), ependymoma (EP) and astrocytoma (PA) is important due to differing medical treatment strategies and predicted survival. The aim of this study was to investigate non-invasive MRI-based radiomic analysis of whole tumors to classify the histologic tumor types of pediatric posterior fossa brain tumor and improve the accuracy of discrimination, using a random forest classifier.MethodsMRI images of 99 patients, with 59 MBs, 13 EPs and 27 PAs histologically confirmed by surgery and pathology before treatment, were included in this retrospective study. Registration was performed between the three sequences, and high- throughput features were extracted from manually segmented tumors on MR images of each case. The forest-based feature selection method was adopted to select the top ten significant features. Finally, the results were compared and analyzed according to the classification.ResultsThe top ten contributions according to the classifier of wavelet features all came from the ADC sequence. The random forest classifier achieved 100% accuracy on the training data and validated the best accuracy (0.938): sensitivity = 1.000, 0.948 and 0.808, specificity = 0.952, 0.926 and 1.000 for EP, MB and PA, respectively.ConclusionA random forest classifier based on the ADC sequence of the whole tumor provides more quantitative information than TIWI and T2WI in differentiating pediatric posterior fossa brain tumors. In particular, the histogram percentile value showed great superiority, which added diagnostic value in pediatric neuro-oncology.     

Focal splenic lesions: Imaging spectrum of diseases on CT,MRI and PET/CT

The spleen can be affected by a variety of diseases. Some of them are readily identified as variations of normal or benign diseases on imaging. However, for a substantial number of focal splenic abnormalities, the diagnosis can be difficult so that histopathologic analysis may be required for a definite diagnosis. In this review, the typical splenic abnormalities that can be diagnosed with imaging with a high degree of confidence are illustrated. The complementary role of computed tomography (CT), magnetic resonance imaging and positron emission tomography/CT that helps make a diagnostic approach is discussed. Finally, current applications and future trends of radiomics and artificial intelligence for the diagnosis of splenic diseases are addressed.