Multitask, Multilabel, and Multidomain Learning With Convolutional Networks for Emotion Recognition.

Automated emotion recognition in the wild from facial images remains a challenging problem. Although recent advances in deep learning have assumed a significant breakthrough in this topic, strong changes in pose, orientation, and point of view severely harm current approaches. In addition, the acquisition of labeled datasets is costly and the current state-of-the-art deep learning algorithms cannot model all the aforementioned difficulties. In this article, we propose applying a multitask learning loss function to share a common feature representation with other related tasks. Particularly, we show that emotion recognition benefits from jointly learning a model with a detector of facial action units (collective muscle movements). The proposed loss function addresses the problem of learning multiple tasks with heterogeneously labeled data, improving previous multitask approaches. We validate the proposal using three datasets acquired in noncontrolled environments, and an application to predict compound facial emotion expressions.

Hypotension in Preterm Infants (HIP) randomised trial.

OBJECTIVE: To determine whether restricting the use of inotrope after diagnosis of low blood pressure (BP) in the first 72 hours of life affects survival without significant brain injury at 36 weeks of postmenstrual age (PMA) in infants born before 28 weeks of gestation. DESIGN: Double-blind, placebo-controlled randomised trial. Caregivers were masked to group assignment. SETTING: 10 sites across Europe and Canada. PARTICIPANTS: Infants born before 28 weeks of gestation were eligible if they had an invasive mean BP less than their gestational age that persisted for ≥15 min in the first 72 hours of life and a cerebral ultrasound free of significant (≥ grade 3) intraventricular haemorrhage. INTERVENTION: Participants were randomly assigned to saline bolus followed by either a dopamine infusion (standard management) or placebo (5% dextrose) infusion (restrictive management). PRIMARY OUTCOME: Survival to 36 weeks of PMA without severe brain injury. RESULTS: The trial terminated early due to significant enrolment issues (7.7% of planned recruitment). 58 infants were enrolled between February 2015 and September 2017. The two groups were well matched for baseline variables. In the standard group, 18/29 (62%) achieved the primary outcome compared with 20/29 (69%) in the restrictive group (p=0.58). Additional treatments for low BP were used less frequently in the standard arm (11/29 (38%) vs 19/29 (66%), p=0.038). CONCLUSION: Though this study lacked power, we did not detect major differences in clinical outcomes between standard or restrictive approach to treatment. These results will inform future studies in this area. TRIAL REGISTRATION NUMBER: NCT01482559, EudraCT 2010-023988-17.

Commentary on the EMA Reflection Paper on the use of extrapolation in the development of medicines for paediatrics.

Adopted guidelines reflect a harmonised European approach to a specific scientific issue and should reflect the most recent scientific knowledge. However, whilst EU regulations are mandatory for all member states and EU directives must be followed by national laws in line with the directive, EMA guidelines do not have legal force and alternative approaches may be taken, but these obviously require more justification. This new series of the BJCP, developed in collaboration with the EMA, aims to address this issue by providing an annotated version of some relevant EMA guidelines and regulatory documents by experts. Hopefully, this will help in promoting their diffusion and in opening a forum for discussion with our readers.

Proposition of a Minimal Effective Dose of Vigabatrin for the Treatment of Infantile Spasms Using Pediatric and Adult Pharmacokinetic Data.

Vigabatrin is an antiepileptic drug indicated as monotherapy in infantile spasms. However, the pharmacokinetic profile of this compound in infants and young children is still poorly understood, as is the minimal effective dose, critical information given the risk of exposure-related retinal toxicity with vigabatrin. A reasonable approach to determining this minimal dose would be to identify the lowest dose providing a low risk of exposure overlap with the 36-mg/kg dose, which is the highest dose associated with an increased risk for treatment failure, based on randomized dose-ranging data. A population pharmacokinetic model was consequently developed from 28 children (aged 0.4-5.7 years) for the active S(+)-enantiomer, using Monolix software. In parallel, a population model was developed from published adult data and scaled to children using theoretical allometry and maturation of the renal function. A one-compartment model with zero-order absorption and first-order elimination described the pediatric data. Mean population estimates (percentage interindividual variability) for the apparent clearance, apparent distribution volume, and absorption duration were 2.36 L/h (24.5%), 17 L (38%), and 0.682 hours, respectively. Apparent clearance and apparent distribution volume were related to body weight by empirical allometric equations. Monte Carlo simulations evidenced that a daily dose of 80 mg/kg should minimize exposure overlap with the 36-mg/kg dose. Similar results were obtained for the adult model scaled to children. Consequently, a minimal effective dose of 80 mg/kg/day could be considered for patients with infantile spasms.

A population pharmacokinetic model taking into account protein binding for the sustained-release granule formulation of valproic acid in children with epilepsy.

PURPOSE: The objective of this work was to develop a population pharmacokinetic model for a prolonged-release granule formulation of valproic acid (VPA) in children with epilepsy and to determine the doses providing a VPA trough concentration (Ctrough) within the target range (50-100 mg/L). METHODS: Ninety-eight children (1-17.6 years, 325 plasma samples) were included in the study. The model was built with NONMEM 7.3. The probability to obtain Ctrough between 50 and 100 mg/L was determined by the Monte Carlo simulations for doses of 20, 30, 40, and 60 mg/kg/day and body weights between 10 and 70 kg. RESULTS: A one compartment model, with first-order absorption and flip-flop parameterization and linear elimination, but taking protein binding into account, was used to describe the data. Typical values for unbound VPA clearance and distribution volume were 6.24 L/h/70 kg and 130 L/h/70 kg respectively. Both parameters were related to body weight via allometric models. The highest probability to obtain a Ctrough within the target range for 10-kg children was obtained with a 40 mg/kg daily dose, whereas daily doses of 30 and 20 mg/kg were found appropriate for 20 to 30- and ≥ 40-kg children respectively. However, for these same doses, the exposure to unbound VPA could differ by 40%. CONCLUSIONS: If the present study supports the current dose recommendations of 20-30 mg/kg/day, except for children under 20 kg, who may need higher doses, it also highlights the need for further research on the pharmacokinetics/pharmacodynamic profile of unbound VPA.

Automated Breast Ultrasound Lesions Detection Using Convolutional Neural Networks.

Breast lesion detection using ultrasound imaging is considered an important step of computer-aided diagnosis systems. Over the past decade, researchers have demonstrated the possibilities to automate the initial lesion detection. However, the lack of a common dataset impedes research when comparing the performance of such algorithms. This paper proposes the use of deep learning approaches for breast ultrasound lesion detection and investigates three different methods: a Patch-based LeNet, a U-Net, and a transfer learning approach with a pretrained FCN-AlexNet. Their performance is compared against four state-of-the-art lesion detection algorithms (i.e., Radial Gradient Index, Multifractal Filtering, Rule-based Region Ranking, and Deformable Part Models). In addition, this paper compares and contrasts two conventional ultrasound image datasets acquired from two different ultrasound systems. Dataset A comprises 306 (60 malignant and 246 benign) images and Dataset B comprises 163 (53 malignant and 110 benign) images. To overcome the lack of public datasets in this domain, Dataset B will be made available for research purposes. The results demonstrate an overall improvement by the deep learning approaches when assessed on both datasets in terms of True Positive Fraction, False Positives per image, and F-measure.

Population Pharmacokinetics of Stiripentol in Paediatric Patients with Dravet Syndrome Treated with Stiripentol, Valproate and Clobazam Combination Therapy.

AIM: The aim of this study was to describe the pharmacokinetics of stiripentol in children with Dravet syndrome and to determine the concentrations of stiripentol achieved in this population for the usual 25 mg/kg twice-daily dose. METHODS: Thirty-five children with epilepsy were included in a prospective population pharmacokinetic study (using MONOLIX software). Four blood samples were drawn per patient. Stiripentol area under the plasma concentration-time curve (AUC) values and trough concentrations were simulated for 7000 theoretical children weighing between 10 and 70 kg for the 25 mg/kg twice-daily dose. RESULTS: The pharmacokinetics of stiripentol was described using a one-compartment model with zero-order absorption and first-order elimination. The apparent clearance (CL/F) and apparent volume of distribution (V d/F) of stiripentol were related to body weight by allometric equations. A dose-dependent non-linearity was also observed with an allometric model relating CL/F to the weight-normalised dose. Mean population estimates (% inter-individual variability) were 4.2 L/h (21%) for CL/F and 82 L (25%) for V d/F. The AUC of stiripentol increased by 300% when body weight increased from 10 to 70 kg. CONCLUSION: This population pharmacokinetic model of stiripentol in children with Dravet syndrome confirmed the dose-dependent non-linearity that has been evidenced in adults. It also supported that a 25 mg/kg twice-daily dose might lead to excessive exposure in children >30 kg, suggesting an eventual dose adjustment during adolescence. CLINICAL TRIAL IDENTIFIER: This study is part of the STIPOP study (EUDRACT number: 2007-001784-30).

Population pharmacokinetics of oxcarbazepine and its monohydroxy derivative in epileptic children.

AIMS: Oxcarbazepine is an antiepileptic drug with an activity mostly due to its monohydroxy derivative metabolite (MHD). A parent-metabolite population pharmacokinetic model in children was developed to evaluate the consistency between the recommended paediatric doses and the reference range for trough concentration (Ctrough ) of MHD (3-35 mg l-1 ). METHODS: A total of 279 plasma samples were obtained from 31 epileptic children (age 2-12 years) after a single dose of oxcarbazepine. Concentration-time data were analysed with Monolix 4.3.2. The probability to obtain Ctrough between 3-35 mg l-1 was determined by Monte Carlo simulations for doses ranging from 10 to 90 mg kg-1  day-1 . RESULTS: A parent-metabolite model with two compartments for oxcarbazepine and one compartment for MHD best described the data. Typical values for oxcarbazepine clearance, central and peripheral distribution volume and distribution clearance were 140 l h-1  70 kg-1 , 337 l 70 kg-1 , 60.7 l and 62.5 l h-1 , respectively. Typical values for MHD clearance and distribution volume were 4.11 l h-1  70 kg-1 and 54.8 l 70 kg-1 respectively. Clearances and distribution volumes of oxcarbazepine and MHD were related to body weight via empirical allometric models. Enzyme-inducing antiepileptic drugs (EIAEDs) increased MHD clearance by 29.3%. Fifty-kg children without EIAEDs may need 20-30 mg kg-1  day-1 instead of the recommended target maintenance dose (30-45 mg kg-1  day-1 ) to obtain Ctrough within the reference range. By contrast, 10-kg children with EIAEDs would need 90 mg kg-1  day-1 instead of the maximum recommended dose of 60 mg kg-1  day-1 . CONCLUSION: This population pharmacokinetic model of oxcarbazepine supports current dose recommendations, except for 10-kg children with concomitant EIAEDs and 50-kg children without EIAEDs.

Commentary on the MID3 Good Practices Paper.

During the last 10 years the European Medicines Agency (EMA) organized a number of workshops on modeling and simulation, working towards greater integration of modeling and simulation (M&S) in the development and regulatory assessment of medicines. In the 2011 EMA - European Federation of Pharmaceutical Industries and Associations (EFPIA) Workshop on Modelling and Simulation, European regulators agreed to the necessity to build expertise to be able to review M&S data provided by companies in their dossier. This led to the establishment of the EMA Modelling and Simulation Working Group (MSWG). Also, there was agreement reached on the need for harmonization on good M&S practices and for continuing dialog across all parties. The MSWG acknowledges the initiative of the EFPIA Model-Informed Drug Discovery and Development (MID3) group in promoting greater consistency in practice, application, and documentation of M&S and considers the paper is an important contribution towards achieving this objective.

National cross-sectional study of nonsteroidal anti-inflammatory drugs use highlights differences between parents and professionals and prompts safety concerns.

AIM: Controversy surrounding the safety of nonsteroidal anti-inflammatory drugs (NSAIDs) provides an opportunity to study parents' and healthcare professionals' differential use of over-the-counter drugs. METHODS: In this national cross-sectional study, general practitioners, paediatricians and pharmacists were asked to include up to five consecutive febrile paediatric patients aged 1 month to 12 years. Parents and healthcare professionals completed questionnaires about the current fever episode. We studied the differential use of NSAIDs by parents and healthcare professionals notably in three clinical conditions with various estimated risk of NSAIDs complications: varicella, gastroenteritis and pharyngitis. RESULTS: The 1534 healthcare professionals prescribed 15% of the 6596 children with an NSAID, but 32% of the parents gave their child an NSAID. Generally, NSAID use was associated with older children, higher temperatures, pain due to otitis and the absence of a rash or gastroenteritis. The differential use of NSAIDs by parents and professionals was greater in conditions with high than low estimated risks of NSAID complications, with odds ratios ranging from to 9.0 to 2.9, respectively. CONCLUSION: The differential use of NSAIDs by healthcare professionals and parents for clinical conditions with potential risks should prompt discussions about the safety of their over-the-counter status.

Enduring large use of acetaminophen suppositories for fever management in children: a national survey of French parents and healthcare professionals' practices.

UNLABELLED: The pharmacological specificities of the rectal formulation of acetaminophen led to a debate on its appropriateness for managing fever in children, but few data are available on the formulation's current use and determinants of use. In a national cross-sectional study between 2007 and 2008, healthcare professionals were asked to include five consecutive patients with acute fever. Among the 6255 children (mean age 4.0 years ± 2.8 SD) who received acetaminophen given by parents or prescribed/recommended by healthcare professionals, determinants of suppository use were studied by multilevel models. A suppository was given by 27 % of parents and prescribed/recommended by 19 % of healthcare professionals, by 24 and 16 %, respectively, for children 2 to 5 years old, and by 13 and 8 %, respectively, for those 6 to 12 years old. Among children who received suppositories from parents and healthcare professionals, 83 and 84 %, respectively, did not vomit. Suppository use was independently associated with several patient- and healthcare professional-level characteristics: young age of children, presence of vomiting, or lack of diarrhea. CONCLUSION: We report an enduring large use of suppositories in France for the symptomatic management of fever in children, including in non-vomiting and/or older children. The rational for such use should be questioned. WHAT IS KNOWN: • The pharmacological specificities of the rectal formulation of acetaminophen have led to a debate on its appropriateness for managing fever in children. Few data are available on the formulation's current use and determinants of the use. What is New: • In a national cross-sectional study, we observed a large use of suppositories in France for symptomatic management of fever in children. Suppositories were frequently used for the youngest children but also for older and/or non-vomiting children.

Ontogeny of ABC and SLC transporters in the microvessels of developing rat brain.

The blood-brain barrier (BBB) is responsible for the control of solutes' concentration in the brain. Tight junctions and multiple ATP-binding cassette (ABC) and SoLute Carrier (SLC) efflux transporters protect brain cells from xenobiotics, therefore reducing brain exposure to intentionally administered drugs. In epilepsy, polymorphisms and overexpression of efflux transporters genes could be associated with pharmacoresistance. The ontogeny of these efflux transporters should also be addressed because their expression during development may be related to different brain exposure to antiepileptic drugs in the immature brain. We detected statistically significant higher expression of Abcb1b and Slc16a1 genes, and lower expression of Abcb1a and Abcg2 genes between the post-natal day 14 (P14) and the adult rat microvessels. P-gP efflux activity was also shown to be lower in P14 rats when compared with the adults. The P-gP proteins coded by rodent genes Abcb1a and Abcb1b are known to have different substrate affinities. The role of the Abcg2 gene is less clear in pharmacoresistance in epilepsy, nonetheless the coded protein Bcrp is frequently associated with drug resistance. Finally, we observed a higher expression of the Mct1 transporter gene in the P14 rat brain microvessels. Accordingly to our results, we suppose that age may be another factor influencing brain exposure to antiepileptics as a consequence of different expression patterns of efflux transporters between the adult and immature BBB.

Pilot evaluation of the population pharmacokinetics of bumetanide in term newborn infants with seizures.

Recent experimental data suggest bumetanide as a possible therapeutic option in newborn infants with seizures after birth asphyxia. Because pharmacokinetic (PK) data are lacking in this population, who very often benefit from therapeutic cooling, which can modify the PK behavior of a drug, a PK study was conducted in term infants with seizures caused by hypoxic-ischemic encephalopathy. Fourteen infants were included, 13 of them being cooled. Forty-nine blood samples were available for the determination of the plasma concentration of bumetanide. Concentration-time data were analyzed by the use of a population approach performed with Monolix Software. Bumetanide was found to follow a 2-compartment model. The mean values were 0.063 L/h for clearance, 0.28 and 0.44 L for the central and peripheral distribution volumes, respectively, and 0.59 L/h for the distribution clearance. Birth body weight explained the interindividual variability of bumetanide clearance via an allometric model. No relationship was found between bumetanide exposure and its efficacy (reduction in seizure burden) or its toxicity (hearing loss). This study describes the first PK model of bumetanide in hypothermia-treated infants with seizures.

Breast-lesion Segmentation Combining B-Mode and Elastography Ultrasound.

Breast ultrasound (BUS) imaging has become a crucial modality, especially for providing a complementary view when other modalities (i.e., mammography) are not conclusive in the task of assessing lesions. The specificity in cancer detection using BUS imaging is low. These false-positive findings often lead to an increase of unnecessary biopsies. In addition, increasing sensitivity is also challenging given that the presence of artifacts in the B-mode ultrasound (US) images can interfere with lesion detection. To deal with these problems and improve diagnosis accuracy, ultrasound elastography was introduced. This paper validates a novel lesion segmentation framework that takes intensity (B-mode) and strain information into account using a Markov Random Field (MRF) and a Maximum a Posteriori (MAP) approach, by applying it to clinical data. A total of 33 images from two different hospitals are used, composed of 14 cancerous and 19 benign lesions. Results show that combining both the B-mode and strain data in a unique framework improves segmentation results for cancerous lesions (Dice Similarity Coefficient of 0.49 using B-mode, while including strain data reaches 0.70), which are difficult images where the lesions appear with blurred and not well-defined boundaries.

Symptomatic Management of Fever in Children: A National Survey of Healthcare Professionals' Practices in France.

Despite the production and dissemination of recommendations related to managing fever in children, this symptom saturates the practices of primary healthcare professionals (HPs). Data on parent practices related to fever are available, but data on HPs' practices are limited. We studied HPs' practices, determinants of practices and concordance with recommendations in France. We conducted a national cross-sectional observational study between 2007 and 2008 among French general practitioners, primary care pediatricians and pharmacists. HPs were asked to include 5 consecutive patients aged 1 month to 12 years with acute fever. HPs completed a questionnaire about their practices for the current fever episode. We used a multilevel logistic regression model to assess the joint effects of patient- and HP-level variables associated with this behavior. In all, 1,534 HPs (participation rate 13%) included 6,596 children (mean age 3.7 ± 2.7 years). Physicians measured the temperature of 40% of children. Primary HPs recommended drug treatment for 84% of children (including monotherapy for 92%) and physical treatment for 62% (including all recommended physical treatments for 7%). HPs gave written advice or a pamphlet for 13% of children. Significant practice variations were associated with characteristics of the child (age, fever level and diagnosis) and HP (profession and experience). In France, despite the production and dissemination of national recommendations for managing fever in children, primary HPs' observed practices differed greatly from current recommendations, which suggests potential targets for continuing medical education.

A review of potential pharmacogenetic effects on catecholamine responses.

Considerably, variability in the clinical response to inotropic agents is observed and could be explained partially by the genetic variants, such as single-nucleotide polymorphism (SNP) in genes encoding for enzymes implicated in catecholamines synthesis, metabolism, storage and release or in the signaling pathway. This review highlights the potential effect of pharmacogenetics studies in hemodynamic response and identified 11 SNPs that could be relevant to explain the high variability drug response for a same dose. Cardiovascular instability, such as hypotension, is one of the premature birth complications. The pharmacogenetics studies evaluating these SNP may be useful to better understand the clinical outcome, particularly in this population.

Bumetanide for the treatment of seizures in newborn babies with hypoxic ischaemic encephalopathy (NEMO): an open-label, dose finding, and feasibility phase 1/2 trial.

BACKGROUND: Preclinical data suggest that the loop-diuretic bumetanide might be an effective treatment for neonatal seizures. We aimed to assess dose and feasibility of intravenous bumetanide as an add-on to phenobarbital for treatment of neonatal seizures. METHODS: In this open-label, dose finding, and feasibility phase 1/2 trial, we recruited full-term infants younger than 48 h who had hypoxic ischaemic encephalopathy and electrographic seizures not responding to a loading-dose of phenobarbital from eight neonatal intensive care units across Europe. Newborn babies were allocated to receive an additional dose of phenobarbital and one of four bumetanide dose levels by use of a bivariate Bayesian sequential dose-escalation design to assess safety and efficacy. We assessed adverse events, pharmacokinetics, and seizure burden during 48 h continuous electroencephalogram (EEG) monitoring. The primary efficacy endpoint was a reduction in electrographic seizure burden of more than 80% without the need for rescue antiepileptic drugs in more than 50% of infants. The trial is registered with ClinicalTrials.gov, number NCT01434225. FINDINGS: Between Sept 1, 2011, and Sept 28, 2013, we screened 30 infants who had electrographic seizures due to hypoxic ischaemic encephalopathy. 14 of these infants (10 boys) were included in the study (dose allocation: 0·05 mg/kg, n=4; 0·1 mg/kg, n=3; 0·2 mg/kg, n=6; 0·3 mg/kg, n=1). All babies received at least one dose of bumetanide with the second dose of phenobarbital; three were withdrawn for reasons unrelated to bumetanide, and one because of dehydration. All but one infant also received aminoglycosides. Five infants met EEG criteria for seizure reduction (one on 0·05 mg/kg, one on 0·1 mg/kg and three on 0·2 mg/kg), and only two did not need rescue antiepileptic drugs (ie, met rescue criteria; one on 0·05 mg/kg and one on 0·3 mg/kg). We recorded no short-term dose-limiting toxic effects, but three of 11 surviving infants had hearing impairment confirmed on auditory testing between 17 and 108 days of age. The most common non-serious adverse reactions were moderate dehydration in one, mild hypotension in seven, and mild to moderate electrolyte disturbances in 12 infants. The trial was stopped early because of serious adverse reactions and limited evidence for seizure reduction. INTERPRETATION: Our findings suggest that bumetanide as an add-on to phenobarbital does not improve seizure control in newborn infants who have hypoxic ischaemic encephalopathy and might increase the risk of hearing loss, highlighting the risks associated with the off-label use of drugs in newborn infants before safety assessment in controlled trials. FUNDING: European Community's Seventh Framework Programme.

Pharmacokinetics of clobazam and N-desmethylclobazam in children with dravet syndrome receiving concomitant stiripentol and valproic Acid.

AIM: The aim of this study was to describe the pharmacokinetics of clobazam and its active metabolite N-desmethylclobazam (N-CLB) in children with Dravet syndrome receiving the stiripentol/valproic acid/clobazam combination therapy of reference and to determine the concentrations of clobazam and N-CLB obtained in this population for the usual 0.2 mg/kg twice-daily dose. METHODS: Thirty-five children with epilepsy were included in a prospective population pharmacokinetic study (using NONMEM(®) software). Four blood samples were drawn per patient. Area under the plasma concentration-time curve (AUC) and trough concentration (C trough) values for clobazam and N-CLB were simulated for 12,000 theoretical children weighing between 10 and 60 kg. RESULTS: The pharmacokinetics of clobazam were described by a one-compartment model with first-order absorption, and elimination, formation and elimination of N-CLB were also first-order processes. The apparent total clearance (CL/F) and distribution volume (V CLB/F) of clobazam and the elimination rate constant of N-CLB (Kem) were related to body weight by allometric equations. Mean population estimates (% inter-individual variability) were 1.23 L/h (29%) for CL/F, 39.1 L (18%) for V CLB/F and 0.0706 h(-1) (26%) for Kem. The AUC values for clobazam and N-CLB were found to increase by 100% when bodyweight increased from 10 to 60 kg, and the simulated C trough values were higher than the currently accepted target values (0.03-0.3 mg/L for clobazam and 0.3-3 mg/L for N-CLB). CONCLUSION: This is the first simultaneous pharmacokinetic model for clobazam and N-CLB in epileptic children. Indicative values for the routine therapeutic drug monitoring of clobazam in children with Dravet syndrome treated by stiripentol are provided. The possible consequences of the weight-related changes on clobazam and N-CLB exposures should be further evaluated.

Computerized detection of breast lesions using deformable part models in ultrasound images.

Ultrasound imaging is considered an important complementary technique for the screening of dense breasts. Detection of lesions at an early stage is a key step in which computerized lesion detection systems could play an important role in the analysis of US images. In this article, we propose adaptation of a generic object detection technique, deformable part models, to detect lesions in breast US images. The data set used in this study included 326 images, all from different patients (54 malignant lesions, 109 benign lesions and 163 healthy breasts). In terms of lesion detection, our proposal outperformed some of the most relevant approaches described in the literature; we obtained a sensitivity of 86% with 0.28 false-positive detection per image and an Az value of 0.975. In the detection of malignant lesions, our proposed approached had an Az value of 0.93 and a sensitivity of 78% at a 1.15 false-positive detections per image.