Encorafenib and binimetinib followed by radiotherapy for patients with BRAFV600-mutant melanoma and brain metastases (E-BRAIN/GEM1802 phase II study).

BACKGROUND: Encorafenib plus binimetinib (EB) is a standard of care treatment for advanced BRAFV600-mutant melanoma. We assessed efficacy and safety of encorafenib plus binimetinib in patients with BRAFV600-mutant melanoma and brain metastasis (BM) and explored if radiotherapy improves the duration of response. METHODS: E-BRAIN/GEM1802 was a prospective, multicenter, single arm, phase II trial that enrolled patients with melanoma BRAFV600-mutant and BM. Patients received encorafenib 450 mg once daily plus binimetinib 45 mg BID, and those who achieved partial response or stable disease at first tumor assessment were offered radiotherapy. Treatment continued until progression.Primary endpoint was intracranial response rate (icRR) after 2 months of EB, establishing a futility threshold of 60%. RESULTS: The study included 25 patients with no BM symptoms and 23 patients with BM symptoms regardless of using corticosteroids. Among them, 31 patients (64.6%) received sequential radiotherapy. After two months, icRR was 70.8% (95% CI: 55.9-83.1); 10.4% complete response. Median intracranial PFS and OS were 8.5 (95% CI: 6.4-11.8) and 15.9 (95% CI: 10.7-21.4) months, respectively (8.3 months for icPFS and 13.9 months OS for patients receiving RDT). Most common grade 3-4 treatment-related adverse event was alanine aminotransferase (ALT) increased (10.4%). CONCLUSION: Encorafenib plus binimetinib showed promising clinical benefit in terms of icRR, and tolerable safety profile with low frequency of high grade TRAEs, in patients with BRAFV600-mutant melanoma and BM, including those with symptoms and need for steroids. Sequential radiotherapy is feasible but it does not seem to prolong response.

Systematic Review of Automated Diuresis Measurement in Critically Ill Patients.

The measurement of urinary flow is a vital medical indicator for critically ill patients in intensive care units. However, there is a clinical need to automate the real-time measurement of diuresis using Internet of Medical Things devices, allowing continuous monitoring of urine flow. A systematic review of scientific literature, patents, and available commercial products was conducted, leading to the conclusion that there is no suitable device to fulfill this need. We identified six characteristics that such a device should possess: minimizing contact with urine, detecting changes in flow patterns, the ability to record minute-by-minute data, capable of sending early alerts, not relying on exclusive disposable components, and being user-friendly for clinical professionals. Additionally, cost-effectiveness is crucial, encompassing the device, infrastructure, maintenance, and usage.

Development of a urinometer for automatic measurement of urine flow in catheterized patients.

Urinary flow measurement and colorimetry are vital medical indicators for critically ill patients in intensive care units. However, there is a clinical need for low-cost, continuous urinary flow monitoring devices that can automatically and in real-time measure urine flow. This need led to the development of a non-invasive device that is easy to use and does not require proprietary disposables. The device operates by detecting urine flow using an infrared barrier that returns an unequivocal pattern, and it is capable of measuring the volume of liquid in real-time, storing the history with a precise date, and returning alarms to detect critical trends. The device also has the ability to detect the color of urine, allowing for extended data and detecting problems in catheterized patients such as hematuria. The device is proposed as an automated clinical decision support system that utilizes the concept of the Internet of Medical Things. It works by using a LoRa communication method with the LoRaWAN protocol to maximize the distance to access points, reducing infrastructure costs in massive deployments. The device can send data wirelessly for remote monitoring and allows for the collection of data on a dashboard in a pseudonymous way. Tests conducted on the device using a gold standard medical grade infusion pump and fluid densities within the 1.005 g/ml to 1.030 g/ml urine density range showed that droplets were satisfactorily captured in the range of flows from less than 1 ml/h to 500 ml/h, which are acceptable ranges for urinary flow. Errors ranged below 15%, when compared to the values obtained by the hospital infusion pump used as gold standard. Such values are clinically adequate to detect changes in diuresis patterns, specially at low urine output ranges, related to renal disfunction. Additionally, tests carried out with different color patterns indicate that it detects different colors of urine with a precision in detecting RGB values <5%. In conclusion, the results suggest that the device can be useful in automatically monitoring diuresis and colorimetry in real-time, which can facilitate the work of nursing and provide automatic decision-making support to intensive care physicians.

Democratization of deep learning for segmenting cartilage from MRIs of human knees: Application to data from the osteoarthritis initiative.

In this study, we aimed to democratize access to convolutional neural networks (CNN) for segmenting cartilage volumes, generating state-of-the-art results for specialized, real-world applications in hospitals and research. Segmentation of cross-sectional and/or longitudinal magnetic resonance (MR) images of articular cartilage facilitates both clinical management of joint damage/disease and fundamental research. Manual delineation of such images is a time-consuming task susceptible to high intra- and interoperator variability and prone to errors. Thus, enabling reliable and efficient analyses of MRIs of cartilage requires automated segmentation of cartilage volumes. Two main limitations arise in the development of hospital- or population-specific deep learning (DL) models for image segmentation: specialized knowledge and specialized hardware. We present a relatively easy and accessible implementation of a DL model to automatically segment MRIs of human knees with state-of-the-art accuracy. In representative examples, we trained CNN models in 6-8 h and obtained results quantitatively comparable to state-of-the-art for every anatomical structure. We established and evaluated our methods using two publicly available MRI data sets originating from the Osteoarthritis Initiative, Stryker Imorphics, and Zuse Institute Berlin (ZIB), as representative test cases. We use Google Colabfor editing and adapting the Python codes and selecting the runtime environment leveraging high-performance graphical processing units. We designed our solution for novice users to apply to any data set with relatively few adaptations requiring only basic programming skills. To facilitate the adoption of our methods, we provide a complete guideline for using our methods and software, as well as the software tools themselves. Clinical significance: We establish and detail methods that clinical personal can apply to create their own DL models without specialized knowledge of DL nor specialized hardware/infrastructure and obtain results comparable with the state-of-the-art to facilitate both clinical management of joint damage/disease and fundamental research.

Advanced Driver Assistance Systems (ADAS) Based on Machine Learning Techniques for the Detection and Transcription of Variable Message Signs on Roads.

Among the reasons for traffic accidents, distractions are the most common. Although there are many traffic signs on the road that contribute to safety, variable message signs (VMSs) require special attention, which is transformed into distraction. ADAS (advanced driver assistance system) devices are advanced systems that perceive the environment and provide assistance to the driver for his comfort or safety. This project aims to develop a prototype of a VMS (variable message sign) reading system using machine learning techniques, which are still not used, especially in this aspect. The assistant consists of two parts: a first one that recognizes the signal on the street and another one that extracts its text and transforms it into speech. For the first one, a set of images were labeled in PASCAL VOC format by manual annotations, scraping and data augmentation. With this dataset, the VMS recognition model was trained, a RetinaNet based off of ResNet50 pretrained on the dataset COCO. Firstly, in the reading process, the images were preprocessed and binarized to achieve the best possible quality. Finally, the extraction was done by the Tesseract OCR model in its 4.0 version, and the speech was done by the cloud service of IBM Watson Text to Speech.

Think Aloud Protocol Applied in Naturalistic Driving for Driving Rules Generation.

Understanding naturalistic driving in complex scenarios is an important step towards autonomous driving, and several approaches have been adopted for modeling driver's behaviors. This paper presents the methodology known as "Think Aloud Protocol" to model driving. This methodology is a data-gathering technique in which drivers are asked to verbalize their thoughts as they are driving which are then recorded, and the ensuing analysis of the audios and videos permits to derive driving rules. The goal of this paper is to show how think aloud methodology is applied in the naturalistic driving area, and to demonstrate the validity of the proposed approach to derive driving rules. The paper presents, firstly, the background of the think aloud methodology and then presents the application of this methodology to driving in roundabouts. The general deployment of this methodology consists of several stages: driver preparation, data collection, audio and video processing, generation of coded transcript files, and the generation of driving rules. The main finding of this study is that think aloud protocol can be applied to naturalistic driving, and even some potential limitations as discussed in the paper, the presented methodology is a relatively easy approach to derive driving rules.

Comparison of machine learning algorithms for clinical event prediction (risk of coronary heart disease).

AIM: The aim of this study is to compare the utility of several supervised machine learning (ML) algorithms for predicting clinical events in terms of their internal validity and accuracy. The results, which were obtained using two statistical software platforms, were also compared. MATERIALS AND METHODS: The data used in this research come from the open database of the Framingham Heart Study, which originated in 1948 in Framingham, Massachusetts as a prospective study of risk factors for cardiovascular disease. Through data mining processes, three data models were elaborated and a comparative methodological study between the different ML algorithms - decision tree, random forest, support vector machines, neural networks, and logistic regression - was carried out. The global selection criterium for choosing the right set of hyperparameters and the type of data manipulation was the area under a curve (AUC). The software tools used to analyze the data were R-Studio® and RapidMiner®. RESULTS: The Framingham study open database contains 4240 observations. The algorithm that yielded the greatest AUC when analyzing the data in R-Studio was neural network applied to a model that excluded all observations in which there was at least one missing value (AUC = 0.71); when analyzing the data in RapidMiner and applying the same model, the best algorithm was support vector machines (AUC = 0.75). CONCLUSIONS: ML algorithms can reinforce the diagnostic and prognostic capacity of traditional regression techniques. Differences between the applicability of those algorithms and the results obtained with them were a function of the software platforms used in the data analysis.

Machine Learning Techniques for Undertaking Roundabouts in Autonomous Driving.

This article presents a machine learning-based technique to build a predictive model and generate rules of action to allow autonomous vehicles to perform roundabout maneuvers. The approach consists of building a predictive model of vehicle speeds and steering angles based on collected data related to driver-vehicle interactions and other aggregated data intrinsic to the traffic environment, such as roundabout geometry and the number of lanes obtained from Open-Street-Maps and offline video processing. The study systematically generates rules of action regarding the vehicle speed and steering angle required for autonomous vehicles to achieve complete roundabout maneuvers. Supervised learning algorithms like the support vector machine, linear regression, and deep learning are used to form the predictive models.

Diagnostic accuracy of ultrasound and fine-needle aspiration in the study of thyroid nodule and multinodular goitre.

OBJECTIVE: Ultrasonography and cytology obtained by fine-needle aspiration are part of the basic study of the thyroid nodule. Although they are not diagnostic in every case, they are cost-effective methods that inform surgical treatment and its extent. The purpose of this study was to evaluate the accuracy of ultrasonography associated with fine-needle aspiration to predict malignancy in nodular thyroid pathology. DESIGN AND PATIENTS: We collected prospective data from patients undergoing thyroidectomy by single nodule or multinodular goitre between 2006 and 2016. A total of 417 patients were included. Ultrasounds were classified as suspected of malignancy if they had 2 or more of the following characteristics: hypoechogenicity, microcalcifications, intranodular central hypervascularization, irregular margins and poorly defined edges. MEASUREMENTS: Ultrasound and fine-needle aspiration accuracy. RESULTS: In the postoperative study, 40% presented malignant pathology. 33% of patients with nonsuspicious ultrasound and 73% of those with suspicious ultrasound had malignant disease. Among patients with single nodule and suspicious ultrasound, the malignancy rate reached 80%. As for cytology, 100% of Bethesda VI patients, 88% of V, 63% of IV, 31% of III and 12% of II were found to have carcinoma. The combination of the 2 tests showed a high predictive value, particularly in cases of Bethesda IV cytology. CONCLUSIONS: Thyroid cytology provides high predictive value of the presence of carcinoma. The predictive value of ultrasound is also high, mainly in the study of isolated nodules. The combination of the 2 tests results in increased diagnostic accuracy.

Fatal gastrointestinal toxicity with ipilimumab after BRAF/MEK inhibitor combination in a melanoma patient achieving pathological complete response.

Approximately 50% of metastatic melanoma patients harbor BRAF mutations. Several treatment options including the combination of BRAF and MEK inhibitors (BRAF/MEKi) and immunotherapy (mainly anti CTLA-4 and anti PD-1 antibodies), have been shown to improve survival in these patients. Although preclinical data support the synergistic effect of both modalities in combination, data confirming the activity and tolerability of these combinations are not yet available in the clinical setting. Herein, we report the case of a melanoma patient treated with sequential BRAF/MEKi (dabrafenib plus trametinib) followed by the anti CTLA-4 antibody ipilimumab who achieved a pathological complete response. Unfortunately, the patient died due to fatal gastrointestinal (GI) toxicity. Analysis of the BRAFV600E mutation in circulating tumoral DNA (ctDNA) from peripheral blood samples and serial tumor tissue biopsies throughout treatment demonstrated a good correlation with clinical evolution.

A dosimetric evaluation of the Eclipse AAA algorithm and Millennium 120 MLC for cranial intensity-modulated radiosurgery.

The aim of this study is to assess the accuracy of a convolution-based algorithm (anisotropic analytical algorithm [AAA]) implemented in the Eclipse planning system for intensity-modulated radiosurgery (IMRS) planning of small cranial targets by using a 5-mm leaf-width multileaf collimator (MLC). Overall, 24 patient-based IMRS plans for cranial lesions of variable size (0.3 to 15.1cc) were planned (Eclipse, AAA, version 10.0.28) using fixed field-based IMRS produced by a Varian linear accelerator equipped with a 120 MLC (5-mm width on central leaves). Plan accuracy was evaluated according to phantom-based measurements performed with radiochromic film (EBT2, ISP, Wayne, NJ). Film 2D dose distributions were performed with the FilmQA Pro software (version 2011, Ashland, OH) by using the triple-channel dosimetry method. Comparison between computed and measured 2D dose distributions was performed using the gamma method (3%/1mm). Performance of the MLC was checked by inspection of the DynaLog files created by the linear accelerator during the delivery of each dynamic field. The absolute difference between the calculated and measured isocenter doses for all the IMRS plans was 2.5% ± 2.1%. The gamma evaluation method resulted in high average passing rates of 98.9% ± 1.4% (red channel) and 98.9% ± 1.5% (blue and green channels). DynaLog file analysis revealed a maximum root mean square error of 0.46mm. According to our results, we conclude that the Eclipse/AAA algorithm provides accurate cranial IMRS dose distributions that may be accurately delivered by a Varian linac equipped with a Millennium 120 MLC.

Feasibility of an online adaptive replanning method for cranial frameless intensity-modulated radiosurgery.

To introduce an approach for online adaptive replanning (i.e., dose-guided radiosurgery) in frameless stereotactic radiosurgery, when a 6-dimensional (6D) robotic couch is not available in the linear accelerator (linac). Cranial radiosurgical treatments are planned in our department using intensity-modulated technique. Patients are immobilized using thermoplastic mask. A cone-beam computed tomography (CBCT) scan is acquired after the initial laser-based patient setup (CBCTsetup). The online adaptive replanning procedure we propose consists of a 6D registration-based mapping of the reference plan onto actual CBCTsetup, followed by a reoptimization of the beam fluences ("6D plan") to achieve similar dosage as originally was intended, while the patient is lying in the linac couch and the original beam arrangement is kept. The goodness of the online adaptive method proposed was retrospectively analyzed for 16 patients with 35 targets treated with CBCT-based frameless intensity modulated technique. Simulation of reference plan onto actual CBCTsetup, according to the 4 degrees of freedom, supported by linac couch was also generated for each case (4D plan). Target coverage (D99%) and conformity index values of 6D and 4D plans were compared with the corresponding values of the reference plans. Although the 4D-based approach does not always assure the target coverage (D99% between 72% and 103%), the proposed online adaptive method gave a perfect coverage in all cases analyzed as well as a similar conformity index value as was planned. Dose-guided radiosurgery approach is effective to assure the dose coverage and conformity of an intracranial target volume, avoiding resetting the patient inside the mask in a "trial and error" way so as to remove the pitch and roll errors when a robotic table is not available.

Dose variations in tumor volumes and organs at risk during IMRT for head-and-neck cancer.

Many head-and-neck cancer (HNC) patients treated with radiotherapy suffer significant anatomical changes due to tumor shrinkage or weight loss. The purpose of this study was to assess dose changes over target volumes and organs at risk during intensity-modulated radiotherapy for HNC patients. Sixteen HNC IMRT patients, all requiring bilateral neck irradiation, were enrolled in the study. A CTplan was performed and the initial dose distribution was calculated. During the treatment, two subsequent CTs at the 15th (CT15) and 25th (CT25) fractions were acquired. The initial plan was calculated on the CT15 and CT25, and dose-volume differences related to the CTplan were assessed. For target volumes, mean values of near-maximun absorbed dose (D2%) increased at the 25th fraction, and doses covering 95% and 98% of volume decreased significantly at the 15th fraction. Contralateral and ipsilateral parotid gland mean doses increased by 6.1% (range: -5.4, 23.5%) and 4.7% (range: -9.1, 22.3%), respectively, at CT25. The D2% in the spinal cord increased by 1.8 Gy at CT15. Mean absorbed dose increases at CT15 and CT25 were observed in: the lips, 3.8% and 5.3%; the oral cavity, 3.5% and 2.5%; and lower middle neck structure, 1.9% and 1.6%. Anatomical changes during treatment of HNC patients affect dose distribution and induce a loss of dose coverage to target volumes and an overdosage to critical structures. Appropriate organs at risk have to be contoured and monitored in order to know if the initial plan remains suitable during the course of the treatment. Reported dosimetric data can help to identify patients who could benefit from adaptive radiotherapy.

Human papillomavirus detection by PCR assay in a large series of high-grade squamous intraepithelial lesions with cytohistological correlation and follow-up.

OBJECTIVE: High-grade squamous intraepithelial lesions (HSIL) are the precursors of invasive cervical carcinomas and are generally associated with the integration of mucosotropic human papillomavirus (HPV) DNA into the host cell genome. Detection of HPV is easy to perform nowadays, even in laboratories with limited technological capacity, and follow-up procedures for patients with HSIL are well established. STUDY DESIGN: HPV detection was performed in a large group of patients with HSIL, and results were correlated with cytological, histological, and colposcopic findings. Discrepancies were examined and discussed. RESULTS: Conventional Papanicolaou (Pap) screening detected 446 HSIL (0.20%) in 218,906 cervical smears. HPV detection by PCR was positive in 339/358 (94.7%) patients. The strains involved were: HPV 16 in 180 patients (53.1%), HPV 18 in 35 (10.3%), HPV 31/33 in 27 (8%), HPV 6/11 in 10 (2.96%), and an unidentified type in 73 (30%). For the last 97 patients (2006-2007), HPV typing was expanded with the following results: HPV 52 was detected in 9 patients (9.2%), HPV 58 in 6 (6.1%), HPV 51 in 4 (4.1%), HPV 68 in 2 (2.0%), and HPV 39 in 1 (1.0%). The number of nonidentified patients dropped to 9 (9.4%); in addition, 14/97 (14.4%) patients were infected with 2 or more viral types. Finally, 19 (5.3%) patients were HPV negative. Colposcopy revealed minor changes in 59 patients (17.3%), major changes in 264 (77.6%), and normal findings in 17 (5.1%). A biopsy was taken in 331/446 patients, and the diagnosis of HSIL or overt malignancy was histologically confirmed in 281 (84.9%) patients: CIN II in 46, CIN III in 224, and histologically upgraded in 11 (6 microinvasive squamous carcinomas, 1 squamous carcinoma, 2 in situ endocervical adenocarcinomas, and 2 microinvasive endocervical adenocarcinomas). Thirty-five patients (10.6%) were downgraded to CIN I and 15 (4.5%) patients had a negative biopsy. Follow-up in the negative-biopsy patients confirmed the existence of SIL in 11 patients [1 HSIL and 10 low-grade squamous intraepithelial lesions (LSIL)] while 4 were considered false positives (atrophic changes, 2; reactive changes, 2). After treatment, 31/331 (9.36%) patients displayed recurrence (HSIL in 29 and LSIL in 2). The viral strains involved in patients with recurrence were HPV 16 in 16 patients (51.6%); HPV 18 in 4 (12.9%); HPV 16 and 18 in 1 (3.2%); HPV 31 in 1 (3.2%); HPV 52 in 1 (3.2%); HPV 18, 31, and 58 in 1 (3.2%); HPV 68 in 1 (3.2%); HPV 51 and 73 in 1 (3.2%), and an unidentified type in 5 (16.1%). Follow-up in 14/19 HSIL and HPV-negative patients confirmed the existence of cervical pathology. CONCLUSIONS: HPV detection improves diagnostic sensitivity and provides an ideal tool for monitoring the response to treatment in HSIL patients. The pathogenic relevance of HPV strain 18 may be greater than previously assumed.

[Locoregional recurrences after conservative treatment of breast cancer stage I-II]. / Recidivas locorregionales tras el tratamiento conservador del cáncer de mama en estadios I y II.

BACKGROUND: Since 1992 conservative treatment of breast cancer (stage I and II: clinical TNM) has been established as an alternative to mastectomy in our hospital. The aim of this retrospective study was to analyse locoregional recurrence features and to compare prognosis with regard to to site of recurrence. METHOD: Between 1987 and 1993, 489 patients with breast cancer (stages I and II) were treated with conservative surgery and radiation therapy at the Hospital Sant Pau of Barcelona. Mean follow-up was 58.8 months [between 12-144]. 35 patients developed locoregional recurrence. We considered two groups: local recurrence in breast only; and locoregional recurrence such as nodal recurrence with or without simultaneous breast recurrence. Diagnosis was confirmed by histopathologic analysis. An extensive study was performed in all patients to rule out distant metastasis. Last follow-up was December 1999. RESULTS: The locoregional recurrence rate after conservative treatment was 7.5% and that of local recurrence was 3.06%. Recurrences were diagnosed in 80% of patients by physical examination, while 20% of patients had noticed the tumor recurrence themselves. Histologic grade III tumors had a higher number of locoregional recurrences than local recurrences (p = 0.030). Locoregional recurrences had lower overall survival rate (p = 0.0005), lower disease-free survival rate (p = 0.0012) and shorter time period without distant metastasis (p < 0.0005) than local recurrences. CONCLUSIONS: Most recurrences were diagnosed by clinical examination during follow up. Histologic grade III was related to locoregional recurrences. Local recurrences had a better prognosis than locoregional recurrences.