Biomedical subjects
Publications and source records attributed to M Juhola.
Medical expert systems are a successful field of applied artificial intelligence. We constructed an otoneurological expert system in our previous research, and in this study we consider its reasoning method. The reasoning process can be described as a modified nearest neighbour solution derived from pattern recognition. The expert system was tested and functions reliably.
Auditory brainstem responses are used to detect hearing defects in audiology and otoneurology. The use of computer programs for the analysis of such recordings is increasing. To identify their detailed properties a pattern recognition algorithm implemented in an analysis program must be highly reliable. For the recognition process, some preprocessing phases after recording the necessary, such as filtering and often also segmentation. In the following, we will explore segmentation, which can be used in preprocessing of biomedical signals after filtering. We studied linear segmentation, where slopes of short signal segments are computed and divided into different classes according to their values. A segment length of 8 samples for a sampling frequency of 50 kHz employed was best according to our tests and error criteria. Using clustering, we found that less than 10 segment classes is suitable for pattern recognition.
We have studied the problem of approximating a digital signal with a suitable continuous broken line. We use the approximative broken line for further analysis of the signal as detection of peaks, waves, and other structural features. We can also save considerable amount of storage space with an approximation that does not lose too much significant information about the original signal. Our work is based on examining different distance metrics and different segmentation methods with respect to the remaining residual error in the resulting approximation. The aim of the work has been to develop a method that can perform segmentation with an acceptable amount of residual error without a need to define a large set of parameters that control the segmentation process. Our contribution is to examine the effect of the estimated compression ratio of the resulting approximation and finding an estimate of this compression ratio. We first define a target in the form of a compression ratio of the resulting approximation and then by applying our method, try to find a suitable threshold parameter to achieve this target. We have tested our method with electrocardiogram (ECG) signals and the compression ratio of the approximation has been found to be a suitable target to control the segmentation process.
An otoneurological expert system was developed to help collect data and diagnose both central and peripheral diseases causing vertigo. Patient history and otoneurological and other examination results are used in the reasoning process. The case history data can be either mandatory or supportive. Mandatory questions are used to confirm a diagnosis, and conflicting answers are used to reject an unlikely disease. Supportive questions support or suppress a diagnosis, but their presence is not obligatory. The reasoning procedure of the otoneurological expert system scores every question independently for different diagnoses, depending on how well they agree with the symptom entity of a disease. Diagnostic criteria are set for each disease. Graphic displays illustrate the linear and nonlinear correlation between the symptoms and diseases. Emphasis is placed on diminishing the possibility of a wrong decision rather than maximizing the likelihood of reaching only one right decision, so that even rare diseases can be taken into consideration.
In the present study linear nonrecursive (FIR) filters designed with the Hamming window and hybrid median filters were examined in connection with computer analysis of sinusoidal tracking eye movements. It is important that digital filters which do not distort sensitive properties of eye movement signals are employed in order to prevent changes caused by too strict filtering in average parameter values of eye movement signals, indicators of some otoneurological disorders. It turned out that within reasonable limits these filters are valid in filtering of sinusoidal tracking eye movement signals.
We investigated two calibration techniques commonly used with eye movement signals pertaining to the vestibulo-ocular reflex. Eye movement signals were recorded electro-oculographically as usual and calibrated using both static and dynamic calibration techniques. The calibration values of normals and patients were computed and compared to each other. Also gain parameters of vestibulo-ocular reflex which depend on the calibration were computed. We found that both techniques are chiefly equally valid, and there are no considerable differences in results computed with either one.
We have developed a Windows-based computer program which will help the user to collect data needed for calculation models of noise induced hearing loss (NIHL). The program has a graphical user interface and it includes several methods to calculate NIHL. We have tried to make our system to cover all the factors concerning NIHL and also to take into account other possible reasons of hearing loss. The system is used not only for estimating noise induced hearing loss, but also as a systematic way to collect data for future evolution of new models and for other research purposes. For this sake the program asks some questions that are not currently included to these models, but which have been shown to have some impact on noise induced hearing loss.
The amplitude ratio and latency between peaks of the auditory brainstem response are widely used as medical parameters of response in the clinical assessment. Several methodological factors affect the medical parameters. The sampling frequency, the resolution of the averager and filtering of the signal are important factors. Typically the signal is highly over-sampled in the recording phase. The resolution of the averager should be as good as possible to ensure adequacy of the amplitude parameter. The latency parameter is more tolerant to the reduction of the sampling frequency, and the signal can be decimated down to 10 kHz to reduce computational complexity.
In this paper, two different otoneurological expert systems, Vertigo and One, the latter developed by us, are considered. The expert systems are evaluated as regards their correctness in reasoning diagnoses. In the light of our data collected from randomly selected test patients, One, being a newer technique, is more effective, since it could infer more cases than vertigo did. All the data was also evaluated and diagnosed by otoneurological specialists, independently of the expert systems, to guarantee objectivity in evaluation of the results of the expert systems.
In this study a computer method is described which is designed to recognize these eye movements originating from vestibulo-ocular reflex signals. The corresponding head movement signal measured concurrently with eye movements serves as the stimulation signal which is utilized in the pattern recognition process and in the computation of gain and phase shift parameters. The recognition method is syntactic with a formal grammar and language applied to terminal symbols into which the measured signals are transformed in a preprocessing phase of the method after filtering. After preliminary tests in an otoneurological laboratory the method seems to function correctly and accurately, and has been taken into use for eye movement investigations of patients suffering from otoneurological diseases.
An interactive database has been developed to assist the diagnostic procedure for vertigo and to store the data. The database offers a possibility to split and reunite the collected information when needed. It contains detailed information about a patient's history, symptoms, and findings in otoneurologic, audiologic, and imaging tests. The symptoms are classified into sets of questions on vertigo (including postural instability), hearing loss and tinnitus, and provoking factors. Confounding disorders are screened. The otoneurologic tests involve saccades, smooth pursuit, posturography, and a caloric test. In addition, findings from specific antibody tests, clinical neurotologic tests, magnetic resonance imaging, brain stem audiometry, and electrocochleography are included. The input information can be applied to workups for vertigo in an expert system called ONE. The database assists its user in that the input of information is easy. If not only can be used for diagnostic purposes but is also beneficial for research, and in combination with the expert system, it provides a tutorial guide for medical students.
Explore the source record for details and available documents.
We studied predictive and pseudo random saccades in patients with cerebellopontine angle tumor. Target for the saccades was a laser beam which was reflected on the wall in front of the subject, and the stimulus was controlled by a computer. The fixed target was spaced at a distance of +/- 30 to each side of the subject. The random saccades jumped 20 degrees/40 degrees/60 degrees in an unpredictable manner. The saccade analysis program was based on pattern recognition. In the random saccade test none of the mean values of saccadic reaction time (SRT), saccadic accuracy (SA) and saccadic peak velocity (SPV) to the right or left differed significantly between the two groups. In the predictable saccade test there was statistically significant differences for certain variables: in right SRT (p < 0.01) and in left SRT (p < 0.01); in right SA (0.001 < p < 0.01) and in left SA (p < 0.001). The mean peak velocity (right SPV and left SPV) did not differ significantly between the two groups. When the stimulus is predictable, the saccades in a healthy subject are more timely spaced and accurate than in a patient. When the stimulus is unpredictable, "the memory of the past" has no importance and cannot help in saccade programming, and variability in the healthy subjects increases.
We have developed an interactive database for vertigo than can be used to assist in the diagnostic procedure and to store the data in a form of a database. The database offers the possibility to split and reunite the collected information in a desired way. The database contains detailed information about patient history, symptoms and findings in neurotological, audiological and imaging tests. The symptoms are classified into three sets of questions: vertigo (including postural instability), hearing loss and tinnitus, and provoking factors. Confounding disorders are screened. The neurotological tests involve saccades, smooth pursuit, posturography and caloric test. In addition, findings in specified antibody testing, clinical neurotological tests. MRI, brain stem audiometry and electrocochleography are included. The input information can be applied in an expert system ONE for vertigo work-up. The database is user-friendly. Besides diagnostic purposes the database is excellent for research purposes, and combined with the expert system it works as a tutorial guide for medical students.
An otoneurological expert system (ONE) was developed to help collect data and diagnose the work-up of vertigo of both central and peripheral diseases causing vertigo. Patient history and otoneurological and other examination results are used in the reasoning process. The history is interactively collected and is complemented with clinical examination results. The case history data can be either mandatory or supportive. Mandatory questions are used to confirm a diagnosis, and conflicting answers are used to reject an unlikely disease. Supportive questions support or suppress a diagnosis, but their presence is not obligatory. The reasoning procedure of ONE scores every question independently for different diagnoses, depending on how well they agree with the symptom entity of a disease. Diagnostic criteria are set for each disease, in Meniere's disease, for example, the full triad is required. Graphic displays illustrate the linear and nonlinear correlation between the symptoms and diseases. For instance, both second-long Tumarkin-type attacks and attacks lasting hours give a high score while intermediately long attacks score much lower in Meniere's disease. To be able to take even rare diseases into consideration we try to diminish the possibility of a wrong decision rather than maximize the likelihood of reaching only one right decision.
Since EOG signals are rather noisy, but magnetic coil signals almost noiseless, the latter type is a good reference to the former which is much easier to record. Concurrently recording magnetic coil and EOG signals we could verify some details as saccades which are often difficult to detect in noisy signals. It is important to recognise such phenomena in eye movement signals, because these affect the gain parameter computed between the vestibulo-ocular reflex signal and the head movement signal.
We examined the pseudo random smooth pursuit test in patients with cerebellopontine angle (CPA) tumors. Four different frequency combinations of sinusoids 0.15-0.35 Hz, 0.25-0.425 Hz, 0.3-0.7 Hz and 0.45-0.7 Hz were tested. Gain was calculated by comparing the smooth pursuit amplitude with stimulus after the supporting saccades had been removed. Phase was calculated by comparing maximum excursion of the eyes and stimulus. Gain was in the patients lower than in the controls in different frequency combinations, and the difference was statistically significant/highly significant in different frequency combinations. In the phase shift the differences were highly significant in the two higher frequency combinations and significant in the second frequency combination (0.25-0.425 Hz), while in the first frequency combination (0.15-0.35 Hz) the difference was not significant. The gain method was a somewhat better analyzing method than the phase shift method.
A replication error (RER) phenotype has been documented both in sporadic colorectal tumors and in tumors from patients with hereditary nonpolyposis colorectal cancer (HNPCC). In the current study 8 of 49 (16%) sporadic colorectal cancers (CRCs) and 25 of 29 (86%) CRCs from HNPCC patients were found to be RER+. All 9 (100%) CRCs from HNPCC patients with germline mutations of the mismatch repair gene MSH2 were found to be RER+, while 16 of 20 CRCs from HNPCC kindreds unlinked or not studied for linkage to MSH2 were RER+. Corresponding analysis in colorectal adenomas revealed that only 1 of 33 (3%) sporadic tumors but 8 of 14 (57%) HNPCC tumors were RER+. Moreover, RER was found in all 6 extracolonic cancers (endometrium, 2; kidney, 1; stomach, 1; duodenum, 1; and ovary, 1) derived from members of HNPCC families. These data suggest the involvement of mismatch repair deficiency in the premalignant stage of tumorigenesis in HNPCC cases, and suggest that mismatch repair genes (MSH2 or others) are defective in the germline of nearly all these patients.