The development and evaluation of CADMIUM: a prototype system to assist in the interpretation of mammograms

OpenClinical.net: Artificial intelligence and knowledge engineering at the point of care

Objective

OpenClinical.net is a way of disseminating clinical guidelines to improve quality of care whose distinctive feature is to combine the benefits of clinical guidelines and other human-readable material with the power of artificial intelligence to give patient-specific recommendations. A key objective is to empower healthcare professionals to author, share, critique, trial and revise these ‘executable’ models of best practice.

Design

OpenClinical.net Alpha (www.openclinical.net) is an operational publishing platform that uses a class of artificial intelligence techniques called knowledge engineering to capture human expertise in decision-making, care planning and other cognitive skills in an intuitive but formal language called PROforma.3 PROforma models can be executed by a computer to yield patient-specific recommendations, explain the reasons and provide supporting evidence on demand.

Results

PROforma has been validated in a wide range of applications in diverse clinical settings and specialties, with trials published in high impact peer-reviewed journals. Trials have included patient workup and risk assessment; decision support (eg, diagnosis, test and treatment selection, prescribing); adaptive care pathways and care planning. The OpenClinical software platform presently supports authoring, testing, sharing and maintenance. OpenClinical’s open-access, open-source repository Repertoire currently carries approximately 50+ diverse examples (https://openclinical.net/index.php?id=69).

Conclusion

OpenClinical.net is a showcase for a PROforma-based approach to improving care quality, safety, efficiency and better patient experience in many kinds of routine clinical practice. This human-centred approach to artificial intelligence will help to ensure that it is developed and used responsibly and in ways that are consistent with professional priorities and public expectations.

Use of a computerised decision aid (DA) to inform the decision process on adjuvant chemotherapy in patients with stage II colorectal cancer: development and preliminary evaluation

OBJECTIVES

To develop a computerised decision aid (DA) to inform the decision process on adjuvant chemotherapy in patients with stage II colorectal cancer, and examine perceived usefulness, acceptability and areas for improvement of the DA.

DESIGN

Mixed methods.

SETTING

Single outpatient oncology department in central London.

PARTICIPANTS

Consecutive recruitment of 13 patients with stage II colorectal cancer, 12 of whom completed the study. Inclusion criteria were: age >18 years; complete resection for stage II adenocarcinoma of the colon or rectum; patients within 14-56 days after surgery; no contraindication to adjuvant chemotherapy; able to give written informed consent. Exclusion criterion: previous chemotherapy.

PRIMARY OUTCOMES

Patient perceived usefulness (assessed by the PrepDM questionnaire) and acceptability of the DA.

RESULTS

PrepDM scores, measuring the perceived usefulness of the DA in preparing the patient to communicate with their doctor and make a health decision, were above those reported in other patient groups. Patient acceptability scores were also high; however, interviews showed that there was evidence of a lack of understanding of key information among some patients, in particular their baseline risk of recurrence, the net benefit of combination chemotherapy and the rationale for having chemotherapy when cancer had apparently gone.

CONCLUSIONS

Patients found the DA acceptable and useful in supporting their decision about whether or not to have adjuvant chemotherapy. Suggested improvements for the DA include: sequential presentation of treatment options (eg, no treatment vs 1 drug, 1 drug vs 2 drugs) to enhance patient understanding of the difference between combination and single therapy, diagrams to help patients understand the rationale for chemotherapy to prevent a recurrence and inbuilt checks on patient understanding of baseline risk of recurrence and net benefit of chemotherapy.

INTEGRATING GLOBAL AND LOCAL ANALYSIS OF COLOR, TEXTURE AND GEOMETRICAL INFORMATION FOR CATEGORIZING LARYNGEAL IMAGES

An approach to integrating the global and local kernel-based automated analysis of vocal fold images aiming to categorize laryngeal diseases is presented in this paper. The problem is treated as an image analysis and recognition task. A committee of support vector machines is employed for performing the categorization of vocal fold images into healthy, diffuse and nodular classes. Analysis of image color distribution, Gabor filtering, cooccurrence matrices, analysis of color edges, image segmentation into homogeneous regions from the image color, texture and geometry view point, analysis of the soft membership of the regions in the decision classes, the kernel principal components based feature extraction are the techniques employed for the global and local analysis of laryngeal images. Bearing in mind the high similarity of the decision classes, the correct classification rate of over 94% obtained when testing the system on 785 vocal fold images is rather encouraging.

Anniversary paper: History and status of CAD and quantitative image analysis: the role of Medical Physics and AAPM

The roles of physicists in medical imaging have expanded over the years, from the study of imaging systems (sources and detectors) and dose to the assessment of image quality and perception, the development of image processing techniques, and the development of image analysis methods to assist in detection and diagnosis. The latter is a natural extension of medical physicists' goals in developing imaging techniques to help physicians acquire diagnostic information and improve clinical decisions. Studies indicate that radiologists do not detect all abnormalities on images that are visible on retrospective review, and they do not always correctly characterize abnormalities that are found. Since the 1950s, the potential use of computers had been considered for analysis of radiographic abnormalities. In the mid-1980s, however, medical physicists and radiologists began major research efforts for computer-aided detection or computer-aided diagnosis (CAD), that is, using the computer output as an aid to radiologists-as opposed to a completely automatic computer interpretation-focusing initially on methods for the detection of lesions on chest radiographs and mammograms. Since then, extensive investigations of computerized image analysis for detection or diagnosis of abnormalities in a variety of 2D and 3D medical images have been conducted. The growth of CAD over the past 20 years has been tremendous-from the early days of time-consuming film digitization and CPU-intensive computations on a limited number of cases to its current status in which developed CAD approaches are evaluated rigorously on large clinically relevant databases. CAD research by medical physicists includes many aspects-collecting relevant normal and pathological cases; developing computer algorithms appropriate for the medical interpretation task including those for segmentation, feature extraction, and classifier design; developing methodology for assessing CAD performance; validating the algorithms using appropriate cases to measure performance and robustness; conducting observer studies with which to evaluate radiologists in the diagnostic task without and with the use of the computer aid; and ultimately assessing performance with a clinical trial. Medical physicists also have an important role in quantitative imaging, by validating the quantitative integrity of scanners and developing imaging techniques, and image analysis tools that extract quantitative data in a more accurate and automated fashion. As imaging systems become more complex and the need for better quantitative information from images grows, the future includes the combined research efforts from physicists working in CAD with those working on quantitative imaging systems to readily yield information on morphology, function, molecular structure, and more-from animal imaging research to clinical patient care. A historical review of CAD and a discussion of challenges for the future are presented here, along with the extension to quantitative image analysis.

A kernel-based approach to categorizing laryngeal images

This paper is concerned with an approach to automated analysis of vocal fold images aiming to categorize laryngeal diseases. Colour, texture, and geometrical features are used to extract relevant information. A committee of support vector machines is then employed for performing the categorization of vocal fold images into healthy, diffuse, and nodular classes. The discrimination power of both, the original and the space obtained based on the kernel principal component analysis is investigated. A correct classification rate of over 92% was obtained when testing the system on 785 vocal fold images. Bearing in mind the high similarity of the decision classes, the correct classification rate obtained is rather encouraging.

Evidence-based guidelines and decision support services: A discussion and evaluation in triple assessment of suspected breast cancer

Widespread health service goals to improve consistency and safety in patient care have prompted considerable investment in the development of evidence-based clinical guidelines. Computerised decision support (CDS) systems have been proposed as a means to implement guidelines in practice. This paper discusses the general concept in oncology and presents an evaluation of a CDS system to support triple assessment (TA) in breast cancer care. Balanced-block crossover experiment and questionnaire study. One stop clinic for symptomatic breast patients. Twenty-four practising breast clinicians from United Kingdom National Health Service hospitals. A web-based CDS system. Clinicians made significantly more deviations from guideline recommendations without decision support (60 out of 120 errors without CDS; 16 out of 120 errors with CDS, P<0.001). Ignoring minor deviations, 16 potentially critical errors arose in the no-decision-support arm of the trial compared with just one (P=0.001) when decision support was available. Opinions of participating clinicians towards the CDS tool became more positive after they had used it (P<0.025). The use of decision support capabilities in TA may yield significant measurable benefits for quality and safety of patient care. This is an important option for improving compliance with evidence-based practice guidelines.

Towards a computer-aided diagnosis system for vocal cord diseases

OBJECTIVE

The objective of this work is to investigate a possibility of creating a computer-aided decision support system for an automated analysis of vocal cord images aiming to categorize diseases of vocal cords.

METHODOLOGY

The problem is treated as a pattern recognition task. To obtain a concise and informative representation of a vocal cord image, colour, texture, and geometrical features are used. The representation is further analyzed by a pattern classifier categorizing the image into healthy, diffuse, and nodular classes.

RESULTS

The approach developed was tested on 785 vocal cord images collected at the Department of Otolaryngology, Kaunas University of Medicine, Lithuania. A correct classification rate of over 87% was obtained when categorizing a set of unseen images into the aforementioned three classes.

CONCLUSION

Bearing in mind the high similarity of the decision classes, the results obtained are rather encouraging and the developed tools could be very helpful for assuring objective analysis of the images of laryngeal diseases.

Characterization of clustered microcalcifications in digitized mammograms using neural networks and support vector machines

OBJECTIVE

Detection and characterization of microcalcification clusters in mammograms is vital in daily clinical practice. The scope of this work is to present a novel computer-based automated method for the characterization of microcalcification clusters in digitized mammograms.

METHODS AND MATERIAL

The proposed method has been implemented in three stages: (a) the cluster detection stage to identify clusters of microcalcifications, (b) the feature extraction stage to compute the important features of each cluster and (c) the classification stage, which provides with the final characterization. In the classification stage, a rule-based system, an artificial neural network (ANN) and a support vector machine (SVM) have been implemented and evaluated using receiver operating characteristic (ROC) analysis. The proposed method was evaluated using the Nijmegen and Mammographic Image Analysis Society (MIAS) mammographic databases. The original feature set was enhanced by the addition of four rule-based features.

RESULTS AND CONCLUSIONS

In the case of Nijmegen dataset, the performance of the SVM was Az=0.79 and 0.77 for the original and enhanced feature set, respectively, while for the MIAS dataset the corresponding characterization scores were Az=0.81 and 0.80. Utilizing neural network classification methodology, the corresponding performance for the Nijmegen dataset was Az=0.70 and 0.76 while for the MIAS dataset it was Az=0.73 and 0.78. Although the obtained high classification performance can be successfully applied to microcalcification clusters characterization, further studies must be carried out for the clinical evaluation of the system using larger datasets. The use of additional features originating either from the image itself (such as cluster location and orientation) or from the patient data may further improve the diagnostic value of the system.

Computer-aided detection in mammography

Mammographic film reading for breast screening is a highly demanding visual task involving a detailed visual search for signs of abnormality, which are infrequent and often small or subtle. False-negative cases, in which a cancer is missed by a film reader, are known to occur. Although double reading has proved effective in reducing errors, there is a national shortage of film readers in the screening programme, and recent extensions to the programme have exacerbated this problem. The use of computer-aided detection (CAD) systems could potentially provide a solution by improving individual performance to the extent that double reading is no longer necessary. In this paper, we describe how CAD works, review the relevant literature and examine the strengths and weaknesses of the approach.

Probability, logic and the cognitive foundations of rational belief

Since Pascal introduced the idea of mathematical probability in the 17th century discussions of uncertainty and “rational” belief have been dogged by philosophical and technical disputes. Furthermore, the last quarter century has seen an explosion of new questions and ideas, stimulated by developments in the computer and cognitive sciences. Competing ideas about probability are often driven by different intuitions about the nature of belief that arise from the needs of different domains (e.g., economics, management theory, engineering, medicine, the life sciences etc). Taking medicine as our focus we develop three lines of argument (historical, practical and cognitive) that suggest that traditional views of probability cannot accommodate all the competing demands and diverse constraints that arise in complex real-world domains. A model of uncertain reasoning based on a form of logical argumentation appears to unify many diverse ideas. The model has precursors in informal discussions of argumentation due to Toulmin, and the notion of logical probability advocated by Keynes, but recent developments in artificial intelligence and cognitive science suggest ways of resolving epistemological and technical issues that they could not address.

Eliciting a terminology for mammographic calcifications

AIM

Two studies were carried out to establish, validate and assess descriptors for use in the differential diagnosis for mammographic calcifications.

METHODS

In Study 1, eleven radiologists were asked to 'think out loud' as they interpreted 20 sets of calcifications. Participants used 159 terms to describe calcifications. We used this data to design a scheme with 50 descriptors. In Study 2, ten radiologists used the scheme to describe 40 sets of calcifications. We assessed the capacity of the terms to discriminate between benign and malignant calcifications, testing them against radiologists' assessments of malignancy and follow-up data.

RESULTS

All descriptors were used by at least 5 radiologists. Five additional descriptors were required. With some exceptions, properties that discriminated between benign and malignant outcomes were highly correlated with radiologists' assessment of risk. Many descriptors have a fairly low sensitivity but high specificity.

CONCLUSIONS

Our data suggest that radiologists consider a wide range of features than is included in existing reporting schemes. Our scheme allows a richer characterization of calcifications, potentially improving the reporting and understanding of these abnormalities.

Decision support and disease management: a logic engineering approach

This paper describes the development and application of PROforma, a unified technology for clinical decision support and disease management. Work leading to the implementation of PROforma has been carried out in a series of projects funded by European agencies over the past 13 years. The work has been based on logic engineering, a distinct design and development methodology that combines concepts from knowledge engineering, logic programming, and software engineering. Several of the projects have used the approach to demonstrate a wide range of applications in primary and specialist care and clinical research. Concurrent academic research projects have provided a sound theoretical basis for the safety-critical elements of the methodology. The principal technical results of the work are the PROforma logic language for defining clinical processes and an associated suite of software tools for delivering applications, such as decision support and disease management procedures. The language supports four standard objects (decisions, plans, actions, and enquiries), each of which has an intuitive meaning with well-understood logical semantics. The development toolset includes a powerful visual programming environment for composing applications from these standard components, for verifying consistency and completeness of the resulting specification and for delivering stand-alone or embeddable applications. Tools and applications that have resulted from the work are described and illustrated, with examples from specialist cancer care and primary care. The results of a number of evaluation activities are included to illustrate the utility of the technology.