Monique Frize

Quantitative assessment of pain-related thermal dysfunction through clinical digital infrared thermal imaging

BackgroundThe skin temperature distribution of a healthy human body exhibits a contralateral symmetry. Some nociceptive and most neuropathic pain pathologies are associated with an alteration of the thermal distribution of the human body. Since the dissipation of heat through the skin occurs for the most part in the form of infrared radiation, infrared thermography is the method of choice to study the physiology of thermoregulation and the thermal dysfunction associated with pain. Assessing thermograms is a complex and subjective task that can be greatly facilitated by computerised techniques.MethodsThis paper presents techniques for automated computerised assessment of thermal images of pain, in order to facilitate the physicians decision making. First, the thermal images are pre-processed to reduce the noise introduced during the initial acquisition and to extract the irrelevant background. Then, potential regions of interest are identified using fixed dermatomal subdivisions of the body, isothermal analysis and segmentation techniques. Finally, we assess the degree of asymmetry between contralateral regions of interest using statistical computations and distance measures between comparable regions.ResultsThe wavelet domain-based Poisson noise removal techniques compared favourably against Wiener and other wavelet-based denoising methods, when qualitative criteria were used. It was shown to improve slightly the subsequent analysis. The automated background removal technique based on thresholding and morphological operations was successful for both noisy and denoised images with a correct removal rate of 85% of the images in the database. The automation of the regions of interest (ROIs) delimitation process was achieved successfully for images with a good contralateral symmetry. Isothermal division complemented well the fixed ROIs division based on dermatomes, giving a more accurate map of potentially abnormal regions. The measure of distance between histograms of comparable ROIs allowed us to increase the sensitivity and specificity rate for the classification of 24 images of pain patients when compared to common statistical comparisons.ConclusionsWe developed a complete set of automated techniques for the computerised assessment of thermal images to assess pain-related thermal dysfunction.

Clinical decision-support systems for intensive care units using case-based reasoning

The artificial intelligence approach used in this work focusses on case-based reasoning techniques for the estimation of medical outcomes and resource utilization. The systems were designed with a view to help medical and nursing personnel to assess patient status, assist in making a diagnosis, and facilitate the selection of a course of therapy. The initial prototype provided information on the closest-matching patient cases to the newest patient admission in an adult intensive care unit (ICU). The system was subsequently re-designed for use in a neonatal ICU. The results of a short clinical pilot evaluation performed in both adult and neonatal units are reported and have led to substantial improvement of the prototype. Future work will include longer-term clinical trials for both adult and neonatal ICUs, once all the software changes have been made to both prototypes in response to the comments of the users made during the preliminary evaluations. To date, the results are very encouraging and physician interest in the potential clinical usefulness of these two systems remains high, and particularly so in the new testing environment in Ottawa.

Design of a health care architecture for medical data interoperability and application integration

The Internet is changing the face of medical research. The current world of isolated research and proprietary data encodings is evolving into a future of standardized medical databases and integrated medical applications, such as clinical decision support systems. This paper explores the use of XML, and its associated Schema Language, to enhance sharing of medical data. XML enables data portability and will reach its full potential when the medical community develops a standardized basis for medical schema content and shares these schemas in recognized repositories. Our research group is currently harnessing XMLs standardization potential by designing a standards-compliant, medical information infrastructure that will allow for seamless integration of all our clinical decision support tools.

Analysis of breast thermography with an artificial neural network

Thermal imaging has been used for early breast cancer detection and risk prediction since the sixties. Examining thermograms for abnormal hyperthermia and hyper-vascularity patterns related to tumor growth is done by comparing images of contralateral breasts. Analysis can be tedious and challenging if the differences are subtle. The advanced computer technology available today can be utilized to automate the analysis and assist in decision-making. In our study, computer routines were used to perform ROI identification and image segmentation of infrared images recorded from 19 patients. Asymmetry analysis between contralateral breasts was carried out to generate statistics that could be used as input parameters to a backpropagation ANN. A simple 1-1-1 network was trained and employed to predict clinical outcomes based on the difference statistics of mean temperature and standard deviation. Results comparing the ANN output with actual clinical diagnosis are presented. Future work will focus on including more patients and more input parameters in the analysis. Performance of ANN network can be studied to select a set of parameters that would best predict the presence of breast cancer.

Weight-elimination neural networks applied to coronary surgery mortality prediction

The objective was to assess the effectiveness of the weight-elimination cost function in improving classification performance of artificial neural networks (ANNs) and to observe how changing the a priori distribution of the training set affects network performance. Backpropagation feedforward ANNs with and without weight-elimination estimated mortality for coronary artery surgery patients. The ANNs were trained and tested on cases with 32 input variables describing the patients medical history; the output variable was in-hospital mortality (mortality rates: training 3.7%, test 3.8%). Artificial training sets with mortality rates of 20%, 50%, and 80% were created to observe the impact of training with a higher-than-normal prevalence. When the results were averaged, weight-elimination networks achieved higher sensitivity rates than those without weight-elimination. Networks trained on higher-than-normal prevalence achieved higher sensitivity rates at the cost of lower specificity and correct classification. The weight-elimination cost function can improve the classification performance when the network is trained with a higher-than-normal prevalence. A network trained with a moderately high artificial mortality rate (artificial mortality rate of 20%) can improve the sensitivity of the model without significantly affecting other aspects of the models performance. The ANN mortality model achieved comparable performance as additive and statistical models for coronary surgery mortality estimation in the literature.

Predicting High-Risk Preterm Birth Using Artificial Neural Networks

A reengineered approach to the early prediction of preterm birth is presented as a complimentary technique to the current procedure of using costly and invasive clinical testing on high-risk maternal populations. Artificial neural networks (ANNs) are employed as a screening tool for preterm birth on a heterogeneous maternal population; risk estimations use obstetrical variables available to physicians before 23 weeks gestation. The objective was to assess if ANNs have a potential use in obstetrical outcome estimations in low-risk maternal populations. The back-propagation feedforward ANN was trained and tested on cases with eight input variables describing the patients obstetrical history; the output variables were: 1) preterm birth; 2) high-risk preterm birth; and 3) a refined high-risk preterm birth outcome excluding all cases where resuscitation was delivered in the form of free flow oxygen. Artificial training sets were created to increase the distribution of the underrepresented class to 20%. Training on the refined high-risk preterm birth model increased the networks sensitivity to 54.8%, compared to just over 20% for the nonartificially distributed preterm birth model

Relative Thresholding With Under-Mattress Pressure Sensors to Detect Central Apnea

Unobtrusive pressure sensors can be used for biological monitoring and long-term health assessment in smart homes. The challenge in detecting events from smart home data is that people have different mattresses, unlike in hospitals where bedding is standardized. This paper proposes to model central apneas using an under-mattress pressure sensor as a measuring instrument. The model uses three parameters, namely, a relative threshold and two time lengths, applied to a moving variance signal. The use of a relative threshold allows apneas to be detected under a variety of different conditions and improves results compared to hard-coded thresholds. The algorithm developed herein was applied to simulated apneas collected from pressure sensors placed under nine different mattresses. The parameters determined from the training set were applied to the test set and produced classification results of 0.78 positive predictive value (PPV) if the bed occupants position is known and 0.75 PPV if the position is unknown. The use of the relative threshold approach overcomes the variability in mattress types found in smart homes.

Preliminary results of severity of illness measures of rheumatoid arthritis using infrared imaging

For the first phase of a large project, we used an infrared imaging camera (thermograph) to obtain accurate measurements of body temperature in joints of twelve human normal subjects (control group) and for thirteen patients who had been diagnosed with rheumatoid arthritis (RA) by a rheumatologist. The ultimate goal is to create a low cost effective method to diagnose early synovitis. Temperature measurements of hands were analyzed with first order statistics. Results show significant temperature differences between control subjects and patients for every joint and hand portion measured. Future work will complete the analysis of knees, elbows, ankles, combine infrared (IR) imaging and intra-optical (IO) imaging, and incorporate feature extraction and classification approaches to stratify patients into severity of illness prior to, and after receiving treatment.

Integrating clinical alerts into an XML-based health care framework for the neonatal intensive care unit

This work extends the functionality of our earlier XML-based health care framework for integrating clinical decision support systems (CDSSs) with capabilities for defining, detecting, and generating clinical alerts in the neonatal intensive care unit (NICU). A first step in this work involved creating a complete NICU XML schema for defining and constraining medical device data, CDSS inputs and outputs, and clinical alerts. The alerts are customizable through a flexible user interface that automatically creates XML documents based on the physicians input specifications. XML documents are transmitted to a central Java application for alert display and transmission. Transmitting XML-based alerts allows the alert information to be shareable in many contexts- within and between hospital information systems, and from remote locations. This is particularly useful when one considers the possibility of offering CDSS-generated alerting systems as ubiquitous web services to pre-authorized users.

Development and usability testing of a parent decision support tool for the neonatal intensive care unit

In this paper we present the development and evaluation of a parent decision support tool for a neonatal intensive care unit (NICU), known as PPADS or Physician and Parent Decision Support. The NICU interprofessional (IP) team uses advanced technology to care for the sickest infants in the hospital, some at the edge of viability. Many difficult care decisions are made daily for this vulnerable population. The PPADS tool, a computerized decision support system, aims to augment current NICU decision-making by helping parents make more informed decisions, improving physician-parent communication, increasing parent decision-making satisfaction, decreasing conflict, and increasing decision efficiency when faced with ethically challenging situations. The development and evaluation of the PPADS tool followed a five step methodology: assessing the clinical environment, establishing the design criteria, developing the system design, implementing the system, and performing usability testing. Usability testing of the PPADS tool with parents of neonates who have graduated (survived) from a tertiary level NICU demonstrates the usefulness and ease of use of the tool.