Deep learning architecture for gesture recognition

Pós-graduação em Engenharia Mecânica - FEG

Social-spider optimization-based artificial neural networks training and its applications for Parkinson's disease identification

Evolutionary algorithms have been widely used for Artificial Neural Networks (ANN) training, being the idea to update the neurons' weights using social dynamics of living organisms in order to decrease the classification error. In this paper, we have introduced Social-Spider Optimization to improve the training phase of ANN with Multilayer perceptrons, and we validated the proposed approach in the context of Parkinson's Disease recognition. The experimental section has been carried out against with five other well-known meta-heuristics techniques, and it has shown SSO can be a suitable approach for ANN-MLP training step.

Optimum-Path Forest Applied for Breast Masses Classification

In Computer-Aided Diagnosis-based schemes in mammography analysis each module is interconnected, which directly affects the system operation as a whole. The identification of mammograms with and without masses is highly needed to reduce the false positive rates regarding the automatic selection of regions of interest for further image segmentation. This study aims to evaluate the performance of three techniques in classifying regions of interest as containing masses or without masses (without clinical findings), as well as the main contribution of this work is to introduce the Optimum-Path Forest (OPF) classifier in this context, which has never been done so far. Thus, we have compared OPF against with two sorts of neural networks in a private dataset composed by 120 images: Radial Basis Function and Multilayer Perceptron (MLP). Texture features have been used for such purpose, and the experiments have demonstrated that MLP networks have been slightly better than OPF, but the former is much faster, which can be a suitable tool for real-time recognition systems.

Extensão do modelo hipercubo para análise de sistemas de atendimento médico emergencial com prioridade na fila

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Cognitive, Mood, and Electroencephalographic Effects of Noninvasive Cortical Stimulation With Weak Electrical Currents

Objectives: The use of noninvasive cortical electrical stimulation with weak currents has significantly increased in basic and clinical human studies. Initial, preliminary studies with this technique have shown encouraging results; however, the safety and tolerability of this method of brain stimulation have not been sufficiently explored yet. The purpose of our study was to assess the effects of direct current (DC) and alternating current (AC) stimulation at different intensities in order to measure their effects on cognition, mood, and electroencephalogram. Methods: Eighty-two healthy, right-handed subjects received active and sham stimulation in a randomized order. We conducted 164 ninety-minute sessions of electrical stimulation in 4 different protocols to assess safety of (1) anodal DC of the dorsolateral prefrontal cortex (DLPFC); (2) cathodal DC of the DLPFC; (3) intermittent anodal DC of the DLPFC and; (4) AC on the zygomatic process. We used weak currents of 1 to 2 mA (for DC experiments) or 0.1 to 0.2 mA (for AC experiment). Results: We found no significant changes in electroencephalogram, cognition, mood, and pain between groups and a low prevalence of mild adverse effects (0.11% and 0.08% in the active and sham stimulation groups, respectively), mainly, sleepiness and mild headache that were equally distributed between groups. Conclusions: Here, we show no neurophysiological or behavioral signs that transcranial DC stimulation or AC stimulation with weak currents induce deleterious changes when comparing active and sham groups. This study provides therefore additional information for researchers and ethics committees, adding important results to the safety pool of studies assessing the effects of cortical stimulation using weak electrical currents. Further studies in patients with neuropsychiatric disorders are warranted.

Using Machine Learning Classifiers to Assist Healthcare-Related Decisions: Classification of Electronic Patient Records

Surveillance Levels (SLs) are categories for medical patients (used in Brazil) that represent different types of medical recommendations. SLs are defined according to risk factors and the medical and developmental history of patients. Each SL is associated with specific educational and clinical measures. The objective of the present paper was to verify computer-aided, automatic assignment of SLs. The present paper proposes a computer-aided approach for automatic recommendation of SLs. The approach is based on the classification of information from patient electronic records. For this purpose, a software architecture composed of three layers was developed. The architecture is formed by a classification layer that includes a linguistic module and machine learning classification modules. The classification layer allows for the use of different classification methods, including the use of preprocessed, normalized language data drawn from the linguistic module. We report the verification and validation of the software architecture in a Brazilian pediatric healthcare institution. The results indicate that selection of attributes can have a great effect on the performance of the system. Nonetheless, our automatic recommendation of surveillance level can still benefit from improvements in processing procedures when the linguistic module is applied prior to classification. Results from our efforts can be applied to different types of medical systems. The results of systems supported by the framework presented in this paper may be used by healthcare and governmental institutions to improve healthcare services in terms of establishing preventive measures and alerting authorities about the possibility of an epidemic.

Monte Carlo dose calculation improvements for low energy electron beams using eMC

The electron Monte Carlo (eMC) dose calculation algorithm in Eclipse (Varian Medical Systems) is based on the macro MC method and is able to predict dose distributions for high energy electron beams with high accuracy. However, there are limitations for low energy electron beams. This work aims to improve the accuracy of the dose calculation using eMC for 4 and 6 MeV electron beams of Varian linear accelerators. Improvements implemented into the eMC include (1) improved determination of the initial electron energy spectrum by increased resolution of mono-energetic depth dose curves used during beam configuration; (2) inclusion of all the scrapers of the applicator in the beam model; (3) reduction of the maximum size of the sphere to be selected within the macro MC transport when the energy of the incident electron is below certain thresholds. The impact of these changes in eMC is investigated by comparing calculated dose distributions for 4 and 6 MeV electron beams at source to surface distance (SSD) of 100 and 110 cm with applicators ranging from 6 x 6 to 25 x 25 cm(2) of a Varian Clinac 2300C/D with the corresponding measurements. Dose differences between calculated and measured absolute depth dose curves are reduced from 6% to less than 1.5% for both energies and all applicators considered at SSD of 100 cm. Using the original eMC implementation, absolute dose profiles at depths of 1 cm, d(max) and R50 in water lead to dose differences of up to 8% for applicators larger than 15 x 15 cm(2) at SSD 100 cm. Those differences are now reduced to less than 2% for all dose profiles investigated when the improved version of eMC is used. At SSD of 110 cm the dose difference for the original eMC version is even more pronounced and can be larger than 10%. Those differences are reduced to within 2% or 2 mm with the improved version of eMC. In this work several enhancements were made in the eMC algorithm leading to significant improvements in the accuracy of the dose calculation for 4 and 6 MeV electron beams of Varian linear accelerators.

Monte Carlo implementation, validation, and characterization of a 120 leaf MLC

Recently, the new high definition multileaf collimator (HD120 MLC) was commercialized by Varian Medical Systems providing high resolution in the center section of the treatment field. The aim of this work is to investigate the characteristics of the HD120 MLC using Monte Carlo (MC) methods.

Generalized eMC implementation for Monte Carlo dose calculation of electron beams from different machine types

The electron Monte Carlo (eMC) dose calculation algorithm available in the Eclipse treatment planning system (Varian Medical Systems) is based on the macro MC method and uses a beam model applicable to Varian linear accelerators. This leads to limitations in accuracy if eMC is applied to non-Varian machines. In this work eMC is generalized to also allow accurate dose calculations for electron beams from Elekta and Siemens accelerators. First, changes made in the previous study to use eMC for low electron beam energies of Varian accelerators are applied. Then, a generalized beam model is developed using a main electron source and a main photon source representing electrons and photons from the scattering foil, respectively, an edge source of electrons, a transmission source of photons and a line source of electrons and photons representing the particles from the scrapers or inserts and head scatter radiation. Regarding the macro MC dose calculation algorithm, the transport code of the secondary particles is improved. The macro MC dose calculations are validated with corresponding dose calculations using EGSnrc in homogeneous and inhomogeneous phantoms. The validation of the generalized eMC is carried out by comparing calculated and measured dose distributions in water for Varian, Elekta and Siemens machines for a variety of beam energies, applicator sizes and SSDs. The comparisons are performed in units of cGy per MU. Overall, a general agreement between calculated and measured dose distributions for all machine types and all combinations of parameters investigated is found to be within 2% or 2 mm. The results of the dose comparisons suggest that the generalized eMC is now suitable to calculate dose distributions for Varian, Elekta and Siemens linear accelerators with sufficient accuracy in the range of the investigated combinations of beam energies, applicator sizes and SSDs.

Can Andean medicine coexist with biomedical healthcare? A comparison of two rural communities in Peru and Bolivia

Background It is commonly assumed that indigenous medical systems remain strong in developing countries because biomedicine is physically inaccessible or financially not affordable. This paper compares the health-seeking behavior of households from rural Andean communities at a Peruvian and a Bolivian study site. The main research question was whether the increased presence of biomedicine led to a displacement of Andean indigenous medical practices or to coexistence of the two healing traditions. Methodology Open-ended interviews and free listing exercises were conducted between June 2006 and December 2008 with 18 households at each study site. Qualitative identification of households’ therapeutic strategies and use of remedies was carried out by means of content analysis of interview transcriptions and inductive interference. Furthermore, a quantitative assessment of the incidence of culture-bound illnesses in local ethnobiological inventories was performed. Results Our findings indicate that the health-seeking behavior of the Andean households in this study is independent of the degree of availability of biomedical facilities in terms of quality of services provided, physical accessibility, and financial affordability, except for specific practices such as childbirth. Preference for natural remedies over pharmaceuticals coexists with biomedical healthcare that is both accessible and affordable. Furthermore, our results show that greater access to biomedicine does not lead to less prevalence of Andean indigenous medical knowledge, as represented by the levels of knowledge about culture-bound illnesses. Conclusions The take-home lesson for health policy-makers from this study is that the main obstacle to use of biomedicine in resource-poor rural areas might not be infrastructural or economic alone. Rather, it may lie in lack of sufficient recognition by biomedical practitioners of the value and importance of indigenous medical systems. We propose that the implementation of health care in indigenous communities be designed as a process of joint development of complementary knowledge and practices from indigenous and biomedical health traditions.

Intrapatient reproducibility of the BISxp monitor

BACKGROUND: The Bispectral Index (BIS) reportedly reflects anesthetic depth. It is recommended that anesthetic agents should be titrated to maintain the BIS between 40 and 60 arbitrary BIS units during anesthesia. For anesthesia providers to follow this recommendation, the monitor should be predictably affected by different anesthetic agents and have good interpatient and intrapatient reproducibility. The authors hypothesized that when two BISxp devices (Aspect Medical Systems, Newton, MA) are placed concurrently on the same patient, their readings are concordant throughout the anesthetic period. METHODS: Simultaneous BIS recordings from two BISxp monitors were obtained during anesthesia at 5-s intervals from 12 participants. RESULTS: In total 22,860 concurrent paired BIS readings were obtained. For 10.7% of the time, there were sustained periods of 30 s or greater where the readings suggested a different depth of anesthesia. For 6% of the time, there were sustained periods of 30 s or greater where the readings differed by 10 or more arbitrary BIS units. The regression coefficient (R) for the two devices was 0.65 (range, 0.35-0.92). There was zero bias between the devices, and the 95% limits of agreement ranged between -18 and +17. CONCLUSION: A conflicting anesthetic management was suggested by the simultaneous BIS readings 10.7% of the time. These results suggest that BISxp does not always provide a reproducible single number. Anesthesia providers should not rely exclusively on the BIS reading when assessing depth of anesthesia.

Introduction of a novel dose saving acquisition mode for the PortalVision aS500 EPID to facilitate on-line patient setup verification

In external beam radiotherapy, electronic portal imaging becomes more and more an indispensable tool for the verification of the patient setup. For the safe clinical introduction of high dose conformal radiotherapy like intensity modulated radiation therapy, on-line patient setup verification is a prerequisite to ensure that the planned dosimetric coverage of the tumor volume is actually realized in the patient. Since the direction of setup fields often deviates from the direction of the treatment beams, extra dose is delivered to the patient during the acquisition of these portal images which may reach clinical relevance. The aim of this work was to develop a new acquisition mode for the PortalVision aS500 electronic portal imaging device from Varian Medical Systems that allows one to take portal images with reduced dose while keeping good image quality. The new acquisition mode, called RadMode, selectively enables and disables beam pulses during image acquisition allowing one to stop wasting valuable dose during the initial acquisition of "reset frames." Images of excellent quality can be taken with 1 MU only. This low dose per image facilitates daily setup verification with considerably reduced extra dose.

Extracorporeal shock wave lithotripsy for distal ureteral calculi: what a powerful machine can achieve

PURPOSE: We assessed the efficacy of extracorporeal shock wave lithotripsy (ESWL, Dornier Medical Systems, Inc., Marietta, Georgia) for distal ureteral calculi with the HM3 (Dornier Medical Systems, Inc.) lithotriptor. MATERIALS AND METHODS: A total of 585 consecutive patients with distal ureteral calculi were treated with ESWL using an unmodified HM3 lithotriptor. Of these patients 67 referred for treatment only for whom no followup was available were excluded from further analysis. The remaining 518 cases were followed until they were radiologically documented to be stone-free or considered treatment failures. Before ESWL additional procedures were performed in 144 patients, including stone push back, ureteral catheter or Double-J (Medical Engineering Corp., New York, New York) stent placement, percutaneous nephrostomy, ureteral endoscopic maneuvers or stone basket manipulation. A total of 374 patients needed no preliminary treatment before ESWL. RESULTS: Of the 518 patients 469 (91%) were successfully treated with 1 ESWL session, while 49 (9%) needed 2 or 3. Manipulation after ESWL was performed in 22 cases, including stent placement, percutaneous nephrostomy, ureteral endoscopic stone removal and a stone basket procedure. On day 1 after ESWL 327 patients (63%) were stone-free, 158 (30%) had less than 5 mm. fragments and 33 (7%) had more than 5 mm. fragments. At 3 months the stone-free rate increased to 97%. CONCLUSIONS: These data show that ESWL for distal ureteral calculi with the powerful unmodified HM3 lithotriptor has a high success rate with a low rate of minimally traumatic manipulations before and after intervention. Results in terms of the re-treatment and stone-free rates are superior to those of any other second or third generation lithotriptor and comparable to the results of the best ureteroscopic series.

New approach to direct stenting using a novel "all-in-one" coronary stent system via 5 French diagnostic catheters: a pilot study

OBJECTIVES We sought to evaluate the strategy success and short term clinical outcomes of direct stenting via 5 French (F) diagnostic catheters using a novel bare metal stent with integrated delivery system (IDS) (Svelte Medical Systems, New Providence, NJ) and compare the results to a conventionally treated matched group. METHODS Fifteen consecutive patients with lesions deemed suitable for direct stenting using a bare metal stent were included. The primary endpoint was the strategy success defined as the ability to successfully treat a target lesion via a 5 F diagnostic catheter with a good angiographic result (TIMI III flow, residual stenosis ≤20%). Procedure and fluoroscopy times, contrast agent use, cost, and short-term clinical outcomes were compared to a matched group treated via conventional stenting. RESULTS The primary endpoint was reached in 14/15 patients (93%). There were no significant differences in procedural (58.6 min ± 12.7 vs. 57.4 min ± 14.2) or fluoroscopy times (10.0 min ± 4.3 vs.10.1 min ± 3.9) or contrast agent use (193.7 ml ± 54.8 vs. 181.4 ml ± 35.6). However, there were significant reductions in materials used in the study group compared to the control group equating to cost savings of almost US $600 per case (US $212.44 ± 258.09 vs. US $804.69 ± 468.11; P = 0.001). CONCLUSIONS Direct stenting using a novel bare metal stent with an IDS via 5 F diagnostic catheters is a viable alternative to conventional stenting in selected patients and is associated with significant cost savings.