White matter tracts and executive functions: a review of causal and correlation evidence.

Executive functions are high-level cognitive processes involving abilities such as working memory/updating, set-shifting and inhibition. These complex cognitive functions are enabled by interactions among widely distributed cognitive networks, supported by white matter tracts. Executive impairment is frequent in neurological conditions affecting white matter; however, whether specific tracts are crucial for normal executive functions is unclear. We review causal and correlation evidence from studies that used direct electrical stimulation during awake surgery for gliomas, voxel-based and tract-based lesion-symptom mapping, and diffusion tensor imaging to explore associations between the integrity of white matter tracts and executive functions in healthy and impaired adults. The corpus callosum was consistently associated with all executive processes, notably its anterior segments. Both causal and correlation evidence showed prominent support of the superior longitudinal fasciculus to executive functions, notably to working memory. More specifically, strong evidence suggested that the second branch of the superior longitudinal fasciculus is crucial for all executive functions, especially for flexibility. Global results showed left lateralization for verbal tasks and right lateralization for executive tasks with visual demands. The frontal aslant tract potentially supports executive functions, however, additional evidence is needed to clarify whether its involvement in executive tasks goes beyond the control of language. Converging evidence indicates that a right-lateralized network of tracts connecting cortical and subcortical grey matter regions supports the performance of tasks assessing response inhibition, some suggesting a role for the right anterior thalamic radiation. Finally, correlation evidence suggests a role for the cingulum bundle in executive functions, especially in tasks assessing inhibition. We discuss these findings in light of current knowledge about the functional role of these tracts, descriptions of the brain networks supporting executive functions and clinical implications for individuals with brain tumours.

Manipulating ΔFOSB in D1-Type Medium Spiny Neurons of the Nucleus Accumbens Reshapes Whole-Brain Functional Connectivity.

BACKGROUND: The transcription factor ΔFOSB, acting in the nucleus accumbens, has been shown to control transcriptional and behavioral responses to opioids and other drugs of abuse. However, circuit-level consequences of ΔFOSB induction on the rest of the brain, which are required for its regulation of complex behavior, remain unknown. METHODS: We used an epigenetic approach in mice to suppress or activate the endogenous Fosb gene and thereby decrease or increase, respectively, levels of ΔFOSB selectively in D1-type medium spiny neurons of the nucleus accumbens and tested whether these modifications affect the organization of functional connectivity (FC) in the brain. We acquired functional magnetic resonance imaging data at rest and in response to a morphine challenge and analyzed both stationary and dynamic FC patterns. RESULTS: The 2 manipulations modified brainwide communication markedly and differently. ΔFOSB down- and upregulation had overlapping effects on prefrontal- and retrosplenial cortex-centered networks, but also generated specific FC signatures for epithalamus (habenula) and dopaminergic/serotonergic centers, respectively. Analysis of dynamic FC patterns showed that increasing ΔFOSB essentially altered responsivity to morphine and uncovered striking modifications of the roles of the epithalamus and amygdala in brain communication, particularly upon ΔFOSB downregulation. CONCLUSIONS: These novel findings illustrate how it is possible to link activity of a transcription factor within a single cell type of an identified brain region to consequent changes in circuit function brainwide by use of functional magnetic resonance imaging, and they pave the way for fundamental advances in bridging the gap between transcriptional and brain connectivity mechanisms underlying opioid addiction.

Volumetric Remodeling of the Left Liver After Right Hepatectomy: Analysis of Factors Predicting Degree of Hypertrophy and Post-hepatectomy Liver Failure.

BACKGROUND: This study investigated the volumetric remodeling of the left liver after right hepatectomy looking for factors predicting the degree of hypertrophy and severe post-hepatectomy liver failure (PHLF). METHODS: In a cohort of 121 right hepatectomies, we performed CT volumetrics study of the future left liver remnant (FLR) preoperatively and postoperatively. Factors influencing FLR degree of hypertrophy and severe PHLF were identified by multivariate analysis. RESULTS: After right hepatectomy, the mean degree of hypertrophy and kinetic growth rate of the left liver remnant were 25% and 3%/day respectively. The mean liver volume recovery rate was 77%. Liver remodeling volume was distributed for 79% on segments 2 and 3 and 21% on the segment 4 (p<0.001). Women showed a greater hypertrophy of segments 2 and 3 compared with men (p=0.002). The degree of hypertrophy of segment 4 was lower in case of middle hepatic vein resection (p=0.004). Left liver remnant kinetic growth rate was associated with the standardized future liver remnant (sFLR) (p<0.001) and a two-stage hepatectomy (p=0.023). Severe PHLF were predicted by intraoperative transfusion (p=0.009), biliary tumors (p=0.013), and male gender (p=0.022). CONCLUSIONS: Volumetric remodeling of the left liver after right hepatectomy is not uniform and is mainly influenced by gender and sacrifice of middle hepatic vein. Male gender, intraoperative transfusion, and biliary tumors increase the risk of postoperative liver failure after right hepatectomy.

Contour-guided deep learning based deformable image registration for dose monitoring during CBCT-guided radiotherapy of prostate cancer.

PURPOSE: To evaluate deep learning (DL)-based deformable image registration (DIR) for dose accumulation during radiotherapy of prostate cancer patients. METHODS AND MATERIALS: Data including 341 CBCTs (209 daily, 132 weekly) and 23 planning CTs from 23 patients was retrospectively analyzed. Anatomical deformation during treatment was estimated using free-form deformation (FFD) method from Elastix and DL-based VoxelMorph approaches. The VoxelMorph method was investigated using anatomical scans (VMorph_Sc) or label images (VMorph_Msk), or the combination of both (VMorph_Sc_Msk). Accumulated doses were compared with the planning dose. RESULTS: The DSC ranges, averaged for prostate, rectum and bladder, were 0.60-0.71, 0.67-0.79, 0.93-0.98, and 0.89-0.96 for the FFD, VMorph_Sc, VMorph_Msk, and VMorph_Sc_Msk methods, respectively. When including both anatomical and label images, VoxelMorph estimated more complex deformations resulting in heterogeneous determinant of Jacobian and higher percentage of deformation vector field (DVF) folding (up to a mean value of 1.90% in the prostate). Large differences were observed between DL-based methods regarding estimation of the accumulated dose, showing systematic overdosage and underdosage of the bladder and rectum, respectively. The difference between planned mean dose and accumulated mean dose with VMorph_Sc_Msk reached a median value of +6.3 Gy for the bladder and -5.1 Gy for the rectum. CONCLUSION: The estimation of the deformations using DL-based approach is feasible for male pelvic anatomy but requires the inclusion of anatomical contours to improve organ correspondence. High variability in the estimation of the accumulated dose depending on the deformable strategy suggests further investigation of DL-based techniques before clinical deployment.

Detecting and quantifying spatial misalignment between longitudinal kilovoltage computed tomography (kVCT) scans of the head and neck by using convolutional neural networks (CNNs).

BACKGROUND: Adaptive radiotherapy (ART) aims to address anatomical modifications appearing during the treatment of patients by modifying the planning treatment according to the daily positioning image. Clinical implementation of ART relies on the quality of the deformable image registration (DIR) algorithms included in the ART workflow. To translate ART into clinical practice, automatic DIR assessment is needed. OBJECTIVE: This article aims to estimate spatial misalignment between two head and neck kilovoltage computed tomography (kVCT) images by using two convolutional neural networks (CNNs). METHODS: The first CNN quantifies misalignments between 0 mm and 15 mm and the second CNN detects and classifies misalignments into two classes (poor alignment and good alignment). Both networks take pairs of patches of 33x33x33 mm3 as inputs and use only the image intensity information. The training dataset was built by deforming kVCT images with basis splines (B-splines) to simulate DIR error maps. The test dataset was built using 2500 landmarks, consisting of hard and soft landmark tissues annotated by 6 clinicians at 10 locations. RESULTS: The quantification CNN reaches a mean error of 1.26 mm (± 1.75 mm) on the landmark set which, depending on the location, has annotation errors between 1 mm and 2 mm. The errors obtained for the quantification network fit the computed interoperator error. The classification network achieves an overall accuracy of 79.32%, and although the classification network overdetects poor alignments, it performs well (i.e., it achieves a rate of 90.4%) in detecting poor alignments when given one. CONCLUSION: The performances of the networks indicate the feasibility of using CNNs for an agnostic and generic approach to misalignment quantification and detection.

Evolution of Neuroimaging Findings in Severe COVID-19 Patients with Initial Neurological Impairment: An Observational Study.

BACKGROUND AND OBJECTIVES: Cerebral complications related to the COVID-19 were documented by brain MRIs during the acute phase. The purpose of the present study was to describe the evolution of these neuroimaging findings (MRI and FDG-PET/CT) and describe the neurocognitive outcomes of these patients. METHODS: During the first wave of the COVID-19 outbreak between 1 March and 31 May 2020, 112 consecutive COVID-19 patients with neurologic manifestations underwent a brain MRI at Strasbourg University hospitals. After recovery, during follow-up, of these 112 patients, 31 (initially hospitalized in intensive care units) underwent additional imaging studies (at least one brain MRI). RESULTS: Twenty-three men (74%) and eight women (26%) with a mean age of 61 years (range: 18-79) were included. Leptomeningeal enhancement, diffuse brain microhemorrhages, acute ischemic strokes, suspicion of cerebral vasculitis, and acute inflammatory demyelinating lesions were described on the initial brain MRIs. During follow-up, the evolution of the leptomeningeal enhancement was discordant, and the cerebral microhemorrhages were stable. We observed normalization of the vessel walls in all patients suspected of cerebral vasculitis. Four patients (13%) demonstrated new complications during follow-up (ischemic strokes, hypoglossal neuritis, marked increase in the white matter FLAIR hyperintensities with presumed vascular origin, and one suspected case of cerebral vasculitis). Concerning the grey matter volumetry, we observed a loss of volume of 3.2% during an average period of approximately five months. During follow-up, the more frequent FDG-PET/CT findings were hypometabolism in temporal and insular regions. CONCLUSION: A minority of initially severe COVID-19 patients demonstrated new complications on their brain MRIs during follow-up after recovery.

An integrative ultrasound-pathology approach to improve preoperative phyllodes tumor classification: A pilot study.

OBJECTIVE: Preoperative diagnosis of phyllodes tumor (PT) is challenging, core-needle biopsy (CNB) has a significant rate of understaging, resulting in suboptimal surgical planification. We hypothesized that the association of imaging data to CNB would improve preoperative diagnostic accuracy compared to biopsy alone. METHODS: In this retrospective pilot study, we included 59 phyllodes tumor with available preoperative imaging, CNB and surgical specimen pathology. RESULTS: Two ultrasound features: tumor heterogeneity and tumor shape were associated with tumor grade, independently of CNB results. Using a machine learning classifier, the association of ultrasound features with CNB results improved accuracy of preoperative tumor classification up to 84%. CONCLUSION: An integrative approach of preoperative diagnosis, associating ultrasound features and CNB, improves preoperative diagnosis and could thus optimize surgical planification.

Predicting pathologic venous invasion before pancreatectomy with venous resection: When does radiology tell the truth?

BACKGROUND: Patients factors in addition to radiological characteristics could predict the presence of pathologic venous invasion in patients undergoing pancreatectomy with venous resection. METHODS: We tested the predictive value of 6 radiological classification methods for predicting pathologic venous invasion-the Nakao, Ishikawa, MD Anderson, Lu, Raptopoulos, and National Comprehensive Cancer Network methods-on a cohort of 198 pancreatectomies (160 pancreaticoduodenectomies and 38 total pancreatectomies) with venous resection for pancreatic adenocarcinomas. Radiological and clinical factors determining pathologic venous invasion were identified by multivariable logistic analysis. RESULTS: Pathologic venous invasion was detected in 124 patients (63.2%). The multivariable logistic regression analysis identified Lu classification (odds ratio = 1.77, 95% confidence interval =1.34-2.35; P < .0001), elevated serum CA19-9 values (odds ratio = 1.97, 95% confidence interval = 1.00-3.90; P = .04), and preoperative neoadjuvant chemotherapy (odds ratio = 0.38, 95% confidence interval = 0.18-0.79; P = .009) as independent factors associated with pathologic venous invasion. Radiological tumor-vessel contact greater than 50% of the circumference or venous wall deformity was associated with a significantly higher rate of pathological venous invasion (80% vs 52%; P < .0001), deeper (media-intima) venous invasion (47% vs 25%; P < .0001), R1 resection (58% vs 41%; P = .03), higher transfusions (84% vs 66%; P = .005), and arterial resection rates (43% vs 27%; P < .0001). Tumor-vein circumference contact of >50% and/or venous wall deformity was still associated with significantly higher rates of pathologic venous invasion, regardless of whether neoadjuvant chemotherapy was used or not and CA19-9 normalized or not under preoperative treatment. CONCLUSION: Preoperative radiological detection of tumor-vein circumference contact >50% and/or venous wall deformity is associated with up to 80% of cases of pathological venous invasion. The combination of radiologic features with biological (CA19-9) and clinical (presence of preoperative chemotherapy) factors could better refine preoperatively the need for venous resection.

Does interventional MRI-guided brain cryotherapy cause a blood-brain barrier disruption? Radiological analysis and perspectives.

Glioblastoma is the most common primary malignant brain tumor with an incidence of 5/100,000 inhabitants/year and a 5-year survival rate of 6.8%. Despite recent advances in the molecular biology understanding of glioblastoma, CNS chemotherapy remains challenging because of the impermeable blood-brain barrier (BBB). Interventional MRI-guided brain cryotherapy (IMRgC) is technique that creates a tissue lesion by making a severe targeted hypothermia and possibly a BBB disruption. This study goal was to analyze the effect of IMRgC on human BBB glioblastoma through its gadolinium enhancing features. All patients harboring a local glioblastoma recurrence and meeting all the inclusion criteria were consecutively included into this retrospective study during a 2-year period. The primary endpoint was to analyze the modification of the gadolinium enhancement on MRI T1 sequences using MR perfusion weighted images during follow-up. The secondary endpoint was to assess any ischemic/hemorrhagic complication following cryotherapy procedure using diffusion weighted imaging (DWI), susceptibility weighted imaging (SWI), or fluid-attenuated inversion recovery (FLAIR). Among the 6 patients studied, all (100%) showed a BBB disruption on the cryotherapy site through the analysis of the perfusion weighted images with an average delay of 2.83 months following the procedure. The gadolinium enhancement located around the cavity then spontaneously decreased in 4/6 patients (67%). No ischemic or hemorrhagic complication was recorded. This study confirms the IMRgC capacity to disrupt BBB as already suggested by the literature. IMRgC might represent a new option in the management of GBM allowing the combined effect of direct cryoablation and enhanced chemotherapy.

Predicting the available space for liver transplantation in cirrhotic patients: a computed tomography-based volumetric study.

BACKGROUND: Anthropometric parameters (weight, height) are usually used for quick matching between two individuals (donor and recipient) in liver transplantation (LT). This study aimed to evaluate clinical factors influencing the overall available space for implanting a liver graft in cirrhotic patients. METHODS: In a cohort of 275 cirrhotic patients undergoing LT, we calculated the liver volume (LV), cavity volume (CV), which is considered the additional space between the liver and the right hypocondrium, and the overall volume (OV = LV + CV) using a computed tomography (CT)-based volumetric system. We then chose the formula based on anthropometric parameters that showed the best predictive value for LV. This formula was used to predict the OV in the same population. Factors influencing OV variations were identified by multivariable logistic analysis. RESULTS: The Hashimoto formula (961.3 × BSA_D-404.8) yielded the lowest median absolute percentage error (21.7%) in predicting the LV. The median LV was 1531 ml. One-hundred eighty-five patients (67.2%) had a median CV of 1156 ml (range: 70-7006), and the median OV was 2240 ml (range: 592-8537). Forty-nine patients (17%) had an OV lower than that predicted by the Hashimoto formula. Independent factors influencing the OV included the number of portosystemic shunts, right anteroposterior abdominal diameter, model for end-stage liver disease (MELD) score > 25, high albumin value, and BMI > 30. CONCLUSIONS: Additional anthropometric characteristics (right anteroposterior diameter, body mass index) clinical (number of portosystemic shunts), and biological (MELD, albumin) factors might influence the overall volume available for liver graft implantation. Knowledge of these factors might be helpful during the donor-recipient matching.

The learning curve for piggyback liver transplantation: identifying factors challenging surgery.

BACKGROUND: This study aimed to quantify the learning curve of piggyback liver transplantation and to identify factors that impact the operative time and blood transfusion during the learning curve. METHODS: A retrospective review was performed on consecutive cases of patients' first piggyback liver transplantations that were performed by a single surgeon. The learning curve for the operative time was evaluated using the cumulative sum method. RESULTS: There were 181, consecutive, first-time piggyback liver transplantations. The median operative time was 345 minutes (range: 180-745 minutes) with a median transfusion rate of 4 packed red blood cell units (range: 0-23 units). The cumulative sum learning curve identified 3 phases: an initial phase (1-70 piggyback liver transplantations), a plateau phase (71-101 piggyback liver transplantations), and a stable phase (102-181 piggyback liver transplantations). Over the 3 phases, there were significant decreases in the median duration of the surgery (388.8 vs 344.8 vs 326.9 minutes; P = .004, P = .0004, P ≤ .0001) and the number of red blood cell units transfused (6.00 vs 3.90 vs 3.71; P = .02, P = .79, P = .0006). Multivariable analysis identified that the following factors impacted the operative time: surgeon experience (P = .00006), previous upper abdominal surgery (P = .01), portocaval shunt fashioning (P = .0003), early portal section (P = .00001), multiple arterial graft reconstruction (P = .03), and the length of the retrohepatic inferior vena covered by segment 1 (P = .0006). Independent risk factors for increased blood loss were surgeon experience (P = .0001), previous upper abdominal surgery (P = .002), the retrohepatic inferior vena cava encirclement by segment 1 (P = .0001), severe portal hypertension (P = .01), early portal section (P = .001), and low prothrombin time (P = .00001). CONCLUSION: Easily identifiable factors related to recipients (segment 1 morphology, previous upper abdominal surgery, severe portal hypertension) and to surgeon (operative experience, portocaval shunt fashioning, early portal section, and multiple arterial reconstructions) impact operative time and blood loss during the learning curve of piggyback liver transplantation. These factors can be used for grading the difficulties of liver transplantation to tailor the surgical strategy.

Large-for-Size Orthotopic Liver Transplantation: a Systematic Review of Definitions, Outcomes, and Solutions.

BACKGROUND: We systematically reviewed the literature on definitions and outcomes of large-for-size (LFS) syndrome in orthotopic liver transplantation (LT). METHODS: This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The Cochrane Library, PubMed, and Embase were searched (January 1990-January 2019) for studies reporting LFS in LT. Primary outcomes were definitions and mortality of LFS LT. RESULTS: Eleven studies reporting patients with LFS LT were identified. Four different formulas (graft-to-recipient weight ratio (GRWR), body surface area index (BSAi), donor standardized total liver volume (sTLV)-to-recipient sTLV ratio, and graft weight/right anteroposterior distance (RAP) ratio) with their critical thresholds were found. There were 81 patients (54% women) with a median weight and height of 62.5 kg (range, 40-105 kg) and 165 cm (range, 145-180 cm). The median graft weight was 1772 g (range, 1290-2400 g), and the median GWRW was 2.77% (range, 2.1-4.00%). Graft venous outflow obstruction was described in seven patients (8.6%). At the time of LT, fascial closure was not achieved in 24 patients (29.6%) and the graft size was reduced by a liver resection in three patients (3.7%). Thirteen deaths (16%) were reported in the first 90 postoperative days with two patients undergoing re-transplant. CONCLUSIONS: LFS LT remains heterogeneously defined but characterized by high mortality rates despite the use of tailored surgical solutions (graft reduction and open abdomen). A composite definition is proposed in order to better describe LFS clinical syndrome.

A Brain Imaging-Based Diagnostic Biomarker for Periodic Catatonia: Preliminary Evidence Using a Bayesian Approach.

Periodic catatonia (PC) is a psychomotor phenotype with a progressive-remitting course. While it can fit any disorder diagnosis of the schizoaffective spectrum, its core features consist of a mix of hypo- and hyperkinesias resulting in distortions of expressive movements such as grimacing and parakinesias. The replication of cerebral blood flow (CBF) increases in the left supplementary motor area (L-SMA) and lateral premotor cortex (L-LPM) in acute and remitting PC patients indicates that these increases could be used as diagnostic biomarkers. In this proof-of-concept study, 2 different MRI sequences were repeated on 3 separate days to get reliable measurement values of CBF in 9 PC and 26 non-PC patients during different cognitive tasks. Each patient was compared to 37 controls. In L-SMA [-9; +10; +60] and L-LPM [-46; -12; +43], a test was positive if the t value was >2.02 (α < 0.05; two tailed). The measurements had good analytical performance. Regarding the tests, their sensitivities and specificities were significantly different from the chance level on both measures, except for L-SMA sensitivities. When combining all the tests, among regions and methods, sensitivity was 98% (95% credible interval [CI] 76-100%) and specificity 88% (72-97%). Bayesian inferences of its negative predictive values for PC were >95% regardless of the context, while its positive predictive values reached 94% but only when used in combination with clinical criteria. The case-by-case analysis suggests that non-PC patients with neurological motor deficits are at risk to be false positive.

Quantitative background parenchymal enhancement to predict recurrence after neoadjuvant chemotherapy for breast cancer.

Breast background parenchymal enhancement (BPE) is an increasingly studied MRI parameter that reflects the microvasculature of normal breast tissue, which has been shown to change during neoadjuvant chemotherapy (NAC) for breast cancer. We aimed at evaluating the BPE in patients undergoing NAC and its prognostic value to predict recurrence. MRI BPE was visually and quantitatively evaluated before and after NAC in a retrospective cohort of 102 women with unilateral biopsy-proven invasive breast cancer. Pre-therapeutic BPE was not predictive of pathological response or recurrence. Quantitative post-therapeutic BPE was significantly decreased compared to pre-therapeutic value. Post-therapeutic quantitative BPE significantly predicted recurrence (HR = 6.38 (0.71, 12.06), p < 0.05).

Evaluation of the effects of chemotherapy on brain glucose metabolism in children with Hodgkin's lymphoma.

OBJECTIVE: Chemobrain is a recently proposed pathological entity. 18F-FDG PET/CT can show objective abnormalities to explain brain disorders caused by chemotherapy, although no study has investigated these phenomena in children to date. The main objective of the present study was to examine quantitatively the effects of chemotherapy on brain metabolism in a homogeneous population of children treated for Hodgkin's lymphoma using 18F-FDG PET/CT. METHODS: In this retrospective study, we included 20 children, newly diagnosed with Hodgkin's lymphoma, who underwent 18F-FDG PET/CT at initial staging and at least one PET/CT in follow-up. The SPM12 software provided t-maps to show the difference in metabolism between these PET/CTs. The statistical maps were analyzed with xjView software to identify the brain regions associated with the clusters detected. RESULTS: Altered glucose metabolism was found in the frontal, cingular, and temporoinsular regions after two cycles of chemotherapy. Results in children were compared to a group of 35 adults. For the same statistical threshold, the extent and depth of the metabolic alterations were less in the adult group than in children. CONCLUSIONS: 18F-FDG PET/CT is useful in providing objective data to explain brain disorders caused by chemotherapy. This could lead to better care and should be compared to neuropsychological test results.

Liver tissue segmentation in multiphase CT scans using cascaded convolutional neural networks.

PURPOSE: We address the automatic segmentation of healthy and cancerous liver tissues (parenchyma, active and necrotic parts of hepatocellular carcinoma (HCC) tumor) on multiphase CT images using a deep learning approach. METHODS: We devise a cascaded convolutional neural network based on the U-Net architecture. Two strategies for dealing with multiphase information are compared: Single-phase images are concatenated in a multi-dimensional features map on the input layer, or output maps are computed independently for each phase before being merged to produce the final segmentation. Each network of the cascade is specialized in the segmentation of a specific tissue. The performances of these networks taken separately and of the cascaded architecture are assessed on both single-phase and on multiphase images. RESULTS: In terms of Dice coefficients, the proposed method is on par with a state-of-the-art method designed for automatic MR image segmentation and outperforms previously used technique for interactive CT image segmentation. We validate the hypothesis that several cascaded specialized networks have a higher prediction accuracy than a single network addressing all tasks simultaneously. Although the portal venous phase alone seems to provide sufficient contrast for discriminating tumors from healthy parenchyma, the multiphase information brings significant improvement for the segmentation of cancerous tissues (active versus necrotic part). CONCLUSION: The proposed cascaded multiphase architecture showed promising performances for the automatic segmentation of liver tissues, allowing to reliably estimate the necrosis rate, a valuable imaging biomarker of the clinical outcome.

Prospective evaluation of ultra-low-dose contrast-enhanced 100-kV abdominal computed tomography with tin filter: effect on radiation dose reduction and image quality with a third-generation dual-source CT system.

OBJECTIVES: To investigate the radiation dose exposure, image quality, and diagnostic performance of enhanced 100-kVp abdominopelvic single-energy CT protocol with tin filter (TF). METHODS: Ninety-three consecutive patients referred for a single-phase enhanced abdominopelvic CT were prospectively included after informed consent. They underwent in addition to a standard protocol (SP) an acquisition with TF. Both examinations were performed on a third-generation dual-source CT system (DSCT), in single energy, using automatic tube current modulation, identical pitch, and identical level of iterative reconstruction. Radiation metrics were compared. Size-specific dose estimates (SSDE), contrast to noise ratio (CNR), and figure of merit (FOM) were calculated. Diagnostic confidence for the assessment of a predetermined list of abdominal lesions was rated by two independent readers. RESULTS: The mean dose of the TF protocol was significantly lower (CDTI 1.56 ± 0.43 mGy vs. 8.13 ± 3.32, p < 0.001; SSDE 9.94 ± 3.08 vs. 1.93 ± 0.39, p < 0.001), with an effective dose close to 1 mSv (1.14 mSv ± 0.34; p < 0.001). TF group exhibited non-significant lower liver CNR (2.76 vs. 3.03, p = 0.56) and was more dose efficient (FOM 10.6 vs. 2.49/mSv, p < 0.001) in comparison to SP. The mean diagnostic confidence for visceral, bone, and peritoneal tumors was equivalent between both groups. CONCLUSIONS: Enhanced 100-kVp abdominopelvic CT acquired after spectral shaping with tin filtration can achieve similar diagnostic performance and CNR compared to a standard CT protocol, while reducing the radiation dose by 81%. KEY POINTS: • 100-kVp spectral filtration enables enhanced abdominal CT with high-dose efficiency. • The radiation dose reaches the 1-mSv range. • Predetermined abdominopelvic lesions can be assessed without impairing on diagnostic confidence.

Survey on deep learning for radiotherapy.

More than 50% of cancer patients are treated with radiotherapy, either exclusively or in combination with other methods. The planning and delivery of radiotherapy treatment is a complex process, but can now be greatly facilitated by artificial intelligence technology. Deep learning is the fastest-growing field in artificial intelligence and has been successfully used in recent years in many domains, including medicine. In this article, we first explain the concept of deep learning, addressing it in the broader context of machine learning. The most common network architectures are presented, with a more specific focus on convolutional neural networks. We then present a review of the published works on deep learning methods that can be applied to radiotherapy, which are classified into seven categories related to the patient workflow, and can provide some insights of potential future applications. We have attempted to make this paper accessible to both radiotherapy and deep learning communities, and hope that it will inspire new collaborations between these two communities to develop dedicated radiotherapy applications.

Automatic detection and segmentation of brain metastases on multimodal MR images with a deep convolutional neural network.

Stereotactic treatments are today the reference techniques for the irradiation of brain metastases in radiotherapy. The dose per fraction is very high, and delivered in small volumes (diameter <1 cm). As part of these treatments, effective detection and precise segmentation of lesions are imperative. Many methods based on deep-learning approaches have been developed for the automatic segmentation of gliomas, but very little for that of brain metastases. We adapted an existing 3D convolutional neural network (DeepMedic) to detect and segment brain metastases on MRI. At first, we sought to adapt the network parameters to brain metastases. We then explored the single or combined use of different MRI modalities, by evaluating network performance in terms of detection and segmentation. We also studied the interest of increasing the database with virtual patients or of using an additional database in which the active parts of the metastases are separated from the necrotic parts. Our results indicated that a deep network approach is promising for the detection and the segmentation of brain metastases on multimodal MRI.