Search

Radiomics, quantitative feature extraction from radiological images, can improve disease diagnosis and prognostication. However, radiomic features are susceptible to image acquisition and segmentation variability. Ideally, only features robust to these variations would be incorporated into predictive models, for good generalisability. We extracted 93 radiomic features from carotid artery computed tomography angiograms of 41 patients with cerebrovascular events. We tested feature robustness to region-of-interest perturbations, image pre-processing settings and quantisation methods using both single- and multi-slice approaches. We assessed the ability of the most robust features to identify culprit and non-culprit arteries using several machine learning algorithms and report the average area under the curve (AUC) from five-fold cross validation. Multi-slice features were superior to single for producing robust radiomic features (67 vs. 61). The optimal image quantisation method used bin widths of 25 or 30. Incorporating our top 10 non-redundant robust radiomics features into ElasticNet achieved an AUC of 0.73 and accuracy of 69% (compared to carotid calcification alone [AUC: 0.44, accuracy: 46%]). Our results provide key information for introducing carotid CT radiomics into clinical practice. If validated prospectively, our robust carotid radiomic set could improve stroke prediction and target therapies to those at highest risk.

Purpose: The aim of the study is to report radiological findings and features in advanced decomposed bodies obtained by post-mortem computed tomography (PMCT) with autopsy correlation. Materials and methods: This retrospective descriptive multicentric study included 41 forensic cases examined between May 2013 and November 2016. All the bodies were PMCT-scanned prior to autopsy, and internal putrefactive state was determined using the radiological alteration index (RAI) by a radiologist with expertise in forensic radiology and a forensic pathologist trained in forensic imaging. After PMCT scans, grade of external putrefaction (GEP) was assigned during the external examination and the complete autopsy was performed by forensic pathologists. Results: The PMCT images evaluation revealed that the RAI index was > 61 in all bodies, corresponding to a moderate-massive presence of putrefactive gas. The gas grade was > II in correspondence of the major vessels, heart cavities, liver parenchyma, vertebra L3 and subcutaneous pectoral tissues, and varied from I to III in correspondence of the kidney. Cadaveric external examination revealed the presence of advanced transformative phenomena, with a GEP3 and GEP4 in most of the cases, with body swelling, eyes and tongue protrusion, body fluids expulsion and fat liquefaction. Conclusion: Radiological imaging by PMCT as an adjunct to autopsy in advanced decomposed bodies represents a useful tool in detecting post-mortem gas, even in very small amounts. A correct interpretation process of the PMCT data is essential to avoid images pitfalls, due to natural decomposition that can be mistaken for pathologic processes.

Purpose: This study was done to assess the diagnostic potential of dual-source computed tomography (DSCT) in the functional evaluation of lung cancer patients undergoing surgical resection. The CT data were compared with pulmonary perfusion scintigraphy and pulmonary function tests (PFTs). Materials and methods: All patients were evaluated with DSCT, scintigraphy and PFTs. The DSCT scan protocol was as follows: two tubes (80 and 140 kV; Care Dose protocol); 70 cc of contrast material (5 cc/s); 5- to 6-s scan time; 0.6 mm collimation. After the automatic calculation of lung perfusion with DSCT and quantification of air volumes and emphysema with dedicated software applications, the perfusional CT studies were compared with scintigraphy using a visual score for perfusion defects; CT air volumes and emphysema were compared with PFTs. Results: The values of accuracy, sensitivity, specificity and positive (PPV) and negative (NPV) predictive values of DSCT compared with perfusion scintigraphy as the reference standard were: 0.88, 0.84, 0.90, 0.93 and 0.88, respectively. The McNemar test did not identify significant differences either between the two imaging techniques (p=0.07) or between CT and PFTs (p=0.09). Conclusions: DSCT is a robust and promising technique that provides important and accurate information on lung function. © 2011 Springer-Verlag Italia.

Purpose: The authors performed a preliminary study with blood-pool contrast-enhanced magnetic resonance angiography (MRA) in evaluating the degree of carotid artery stenosis and plaque morphology, comparing the diagnostic performance of first-pass (FP) and steady-state (SS) acquisitions with 64-slice computed tomography angiography (CTA) and using digital subtraction angiography (DSA) as the reference standard. Materials and methods: Twenty patients with ≥50% carotid artery stenosis at Doppler sonography underwent blood-pool contrast-enhanced MRA, CTA and DSA. Two independent radiologists evaluated MRA and CTA examinations to assess the degree of stenosis and characterise plaque morphology. Accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated for FP, SS and CTA. The McNemar and Wilcoxon tests were used to determine significant differences (p0.05). Conclusions: Blood-pool contrast-enhanced MRA with SS sequences allow improved diagnostic evaluation of the degree of carotid stenosis and plaque morphology compared with FP and is substantially equal to CTA and DSA. © 2010 Springer-Verlag Italia Riassunto.

Purpose. The aim of this study was to validate a 64-row multidetector computed tomography (64-MDCT) acquisition protocol with biphasic administration of contrast medium for comprehensive assessment of the coronary and systemic arterial tree in a single examination. Materials and methods. The scanning protocol comprised two acquisitions: an electrocardiograph (ECG)-gated scan at the level of the heart, followed by a total-body, low-dose scan of the systemic arterial circulation. Twenty patients were evaluated using two different strategies for contrast administration. In ten patients, the delay between the two acquisitions was set at 40 s, whereas in the remaining patients, it varied between 45 s and 65 s. For both strategies, the degree of systemic arterial opacification and the attenuation gradient between arterial and venous structures were quantitatively assessed at six extracoronary locations. Two observers evaluated in consensus the presence or absence of atherosclerosis and the degree of stenosis of arterial segments. Results. Three hundred coronary segments were analysed. Arterial-wall changes were depicted in 155 (51%) segments, and in 35 (23%), the degree of stenosis was >50%. Of the 640 extracoronary arterial segments, 250 (39%) presented atherosclerotic wall alterations, in 50 (20%), the degree of stenosis was >50% and five were affected by aneurysmal dilatation. The magnitude of arterial opacification values and attenuation gradients between arterial and venous structures were significantly higher in patients scanned with the 40-s fixed-delay strategy. Conclusions. Whole-body CT angiography with biphasic administration of contrast agent and fixed scan delay has been shown to be a feasible and reproducible technique. Comprehensive data on the global atherosclerotic burden potentially offer important therapeutic options for subclinical, high-risk segments. © 2008 Springer-Verlag.

Hereditary breast cancer accounts for about 10% of all breast cancers and BRCA1 and BRCA2 genes have been identified as validated susceptibility genes for this pathology. Testing for BRCA gene mutations is usually based on a pre-screening approach, such as the partial denaturation DHPLC method, and capillary direct sequencing. However, this approach is time consuming due to the large size of BRCA1 and BRCA2 genes. Recently, a new low cost and time saving DHPLC protocol has been developed to analyze gene mutations by using SURVEYOR(®) Nuclease digestion and DHPLC analysis. A subset of 90 patients, enrolled in the Genetic Counseling Program of the National Cancer Centre of Bari (Italy), was performed to validate this approach. Previous retrospective analysis showed that 9/90 patients (10%) were mutated in BRCA1 and BRCA2 genes and these data were confirmed by the present approach. DNA samples underwent touchdown PCR and, subsequently, SURVEYOR(®) nuclease digestion. BRCA1 and BRCA2 amplicons were divided into groups depending on amplicon size to allow multiamplicon digestion. The product of this reaction were analyzed on Transgenomic WAVE Nucleic Acid High Sensitivity Fragment Analysis System. The operator who performed the DHPLC surveyor approach did not know the sequencing results at that time. The SURVEYOR(®) Nuclease DHPLC approach was able to detect all alterations with a sensitivity of 95%. Furthermore, in order to save time and reagents, a multiamplicon setting preparation was validated.

To assess safety and effectiveness of non-invasive high intensity 3T MR guided Focused Ultrasound Surgery (MRgFUS) treatment of localized prostate cancer as salvage therapy and as an alternative to active surveillance.

Background: Current treatment for melanoma of the lower limb includes excision of the primary tumor with ilioinguinal lymphadenectomy in the case of lymph node metastases. The standard surgical approach includes sectioning of the inguinal ligament to gain access to the iliac nodes. More recently, some authors have reported that extraperitoneal laparoscopically assisted ilioinguinal lymphadenectomy for the treatment of malignant melanoma is feasible and less aggressive than standard open surgery. So far, no publications have described transperitoneal laparoscopic iliac lymphadenectomy (TPLND). Methods: From November 2001 to June 2002, 13 patients with ilioinguinal node melanoma metastases underwent TPLND (stage IIIA in 1 case, IIIB in 5 cases, IIIC in 4 cases, and IV in 3 cases). Results: In all 13 cases, the TPLND and groin dissection was performed correctly. Operative time, intra- and postoperative complications, number of lymph nodes retrieved, immediate morbidity, hospital stay, and feasibility of TPLND were evaluated. Conclusions: This study was conducted to evaluate the feasibility and the preliminary results of TPLND used to manage malignant melanoma of the lower limb. This approach has many advantages over the traditional procedure: less surgical trauma, no incision of the abdominal muscles or the inguinal ligament, and less postoperative pain. Moreover, as compared with extraperitoneal laparoscopically assisted ilioinguinal lymphoadenectomy, it provides an improved view of the operative area, dissection zone, and surrounding structures. Further research is needed to confirm these preliminary results regarding the potential applications of this method for treating malignant metastasis to the lower limb.

Materials and methods After giving their informed consent 5 patients with histologically proven unresectable pancreatic adenocarcinoma and 1 patient with unresectable right lobe HCC (4 males, 2 females; age range 58-72) underwent MRgFUS treatment on a dedicated 3T unit featuring the ExAblate 2100 system (InSightec). The system is composed by a 200-element transducer located within the MR table. The MR guidance allows a detailed depiction and visualization of the lesion; moreover, the use of the proton resonance frequency (PRF) shift method allows a real time monitoring of the temperature inside the target lesion and the adjacent anatomical structures, in order to ensure adequate tissue ablation and safe ablation margins. The treatment was performed in general anesthesia with breath control. After the procedure, gadolinium-enhanced gradient echo T1-weighted sequences were performed in order to evaluate the ablated area and the absence of possible local complications. Clinical and imaging follow-up was performed with both MR and CT at 3 and 6 months after treatment respectively for the patient with HCC and those with pancreatic cancer. Results Treatment was successfully performed in all patients without any adverse events during or after the procedure. MR images acquired immediately after treatment demonstrated necrosis of ablated area within the lesion in all cases; in particular the HCC was completely non-enhancing. At short term clinical follow-up, all the patients with pancreatic cancer referred reduction of pain symptoms due to infiltration of the celiac plexus. However follow-up imaging demonstrated recurrence of pathologic tissue within the ablated area, even if there was no local progression of the disease. Two patients with pancreatic cancer underwent radiotherapy after treatment, while the remaining one underwent another MRgFUS ablation.

Materials and method 110 leiomyomas in 98 women (average age 39,3 years) were treated with MRI-guided focused ultrasound (MRgFUS). The treatment is carried out using the ablative properties of the HIFU system under 3T MRI guide. Symptoms (e.g. menorrhagia, pelvic pain) were scored using Severity Score (SS) and quality of life was determined using the UFS-QOL score. Pre-treatment measurementes of leiomyoma volume were obtained by MR images. Immediately after treatment, Non-perfused Volume (NPV) was calculated from T1-w contrastenhanced MR sequences. The average volume of treated fibroids was 90.27 ± 90.4 mm3. Follow-up images were obtained at 3 and 12 months after treatment and served to determine leiomyoma shrinkage. Qualitative and quantitative relations between fibroid volume, NPV ratio at treatment, and 12-month shrinkage were measured.

Radiomics, quantitative feature extraction from radiological images, can improve disease diagnosis and prognostication. However, radiomic features are susceptible to image acquisition and segmentation variability. Ideally, only features robust to these variations would be incorporated into predictive models, for good generalisability. We extracted 93 radiomic features from carotid artery computed tomography angiograms of 41 patients with cerebrovascular events. We tested feature robustness to region-of-interest perturbations, image pre-processing settings and quantisation methods using both single- and multi-slice approaches. We assessed the ability of the most robust features to identify culprit and non-culprit arteries using several machine learning algorithms and report the average area under the curve (AUC) from five-fold cross validation. Multi-slice features were superior to single for producing robust radiomic features (67 vs. 61). The optimal image quantisation method used bin widths of 25 or 30. Incorporating our top 10 non-redundant robust radiomics features into ElasticNet achieved an AUC of 0.73 and accuracy of 69% (compared to carotid calcification alone [AUC: 0.44, accuracy: 46%]). Our results provide key information for introducing carotid CT radiomics into clinical practice. If validated prospectively, our robust carotid radiomic set could improve stroke prediction and target therapies to those at highest risk.