Keratoconus detection with the KISA% method-another view

Keratoconus Diagnosis: From Fundamentals to Artificial Intelligence: A Systematic Narrative Review

The remarkable recent advances in managing keratoconus, the most common corneal ectasia, encouraged researchers to conduct further studies on the disease. Despite the abundance of information about keratoconus, debates persist regarding the detection of mild cases. Early detection plays a crucial role in facilitating less invasive treatments. This review encompasses corneal data ranging from the basic sciences to the application of artificial intelligence in keratoconus patients. Diagnostic systems utilize automated decision trees, support vector machines, and various types of neural networks, incorporating input from various corneal imaging equipment. Although the integration of artificial intelligence techniques into corneal imaging devices may take time, their popularity in clinical practice is increasing. Most of the studies reviewed herein demonstrate a high discriminatory power between normal and keratoconus cases, with a relatively lower discriminatory power for subclinical keratoconus.

Anterior and posterior corneal higher-order aberrations in early diagnosis and grading of keratoconus

CLINICAL RELEVANCE

Evaluation of corneal higher-order aberrations can be used clinically to diagnose early cases of keratoconus as well as to classify the severity of keratoconus.

BACKGROUND

To investigate the anterior and posterior corneal higher-order aberrations (HOAs) up to the sixth order and their ability to identify early keratoconus (KCN) as well as differentiate different severities of KCN using cross-validation analysis.

METHODS

This prospective cross-sectional comparative study was performed at a tertiary eye hospital in Tehran, Iran, in 2019. The study sample consisted of 95 eyes of 95 patients with KCN and 53 eyes of 53 normal individuals. The eyes with KCN were classified into three groups based on the Amsler-Krumeich classification system: group 1 (mild KCN), group 2 (moderate KCN), and group 3 (severe KCN). Corneal wavefront analysis was performed using Pentacam HR.

RESULTS

Based on the magnitude of AUC, posterior vertical secondary coma (Z5^-1) had an excellent discriminant ability (AUC: 0.91) and anterior vertical coma (Z3^-1) and anterior vertical secondary coma (Z5^-1) had a good discriminant ability (0.8 < AUC < 0.89) for differentiating eyes with mild KCN from normal eyes. The anterior and posterior primary spherical aberrations (Z4°) had an excellent ability (AUC > 0.9), and anterior secondary spherical aberration (Z6°) had a good ability (AUC: 0.83) for differentiating moderate from mild KCN. In the differentiation of severe from moderate KCN, anterior and posterior primary aspherical aberrations (Z4°) had a good AUC value (AUC > 0.8).

CONCLUSION

Coma-like aberrations had a good discriminant ability between normal eyes and eyes with mild KCN. Spherical aberrations showed a good ability for differentiating between different stages of KCN. The cut-off values reported in this study can be used for early detection of KCN as well as classification of KCN severity.

[Keratoconus detection and classification from parameters of the Corvis®ST : A study based on algorithms of machine learning]

BACKGROUND AND OBJECTIVE

In the last decades increasingly more systems of artificial intelligence have been established in medicine, which identify diseases or pathologies or discriminate them from complimentary diseases. Up to now the Corvis®ST (Corneal Visualization Scheimpflug Technology, Corvis®ST, Oculus, Wetzlar, Germany) yielded a binary index for classifying keratoconus but did not enable staging. The purpose of this study was to develop a prediction model, which mimics the topographic keratoconus classification index (TKC) of the Pentacam high resolution (HR, Oculus) with measurement parameters extracted from the Corvis®ST.

PATIENTS AND METHODS

In this study 60 measurements from normal subjects (TKC 0) and 379 eyes with keratoconus (TKC 1-4) were recruited. After measurement with the Pentacam HR (target parameter TKC) a measurement with the Corvis®ST device was performed. From this device 6 dynamic response parameters were extracted, which were included in the Corvis biomechanical index (CBI) provided by the Corvis®ST (ARTh, SP-A1, DA ratio 1 mm, DA ratio 2 mm, A1 velocity, max. deformation amplitude). In addition to the TKC as the target, the binarized TKC (1: TKC 1-4, 0: TKC 0) was modelled. The performance of the model was validated with accuracy as an indicator for correct classification made by the algorithm. Misclassifications in the modelling were penalized by the number of stages of deviation between the modelled and measured TKC values.

RESULTS

A total of 24 different models of supervised machine learning from 6 different families were tested. For modelling of the TKC stages 0-4, the algorithm based on a support vector machine (SVM) with linear kernel showed the best performance with an accuracy of 65.1% correct classifications. For modelling of binarized TKC, a decision tree with a coarse resolution showed a superior performance with an accuracy of 95.2% correct classifications followed by the SVM with linear or quadratic kernel and a nearest neighborhood classifier with cubic kernel (94.5% each).

CONCLUSION

This study aimed to show the principle of supervised machine learning applied to a set-up for the modelled classification of keratoconus staging. Preprocessed measurement data extracted from the Corvis®ST device were used to mimic the TKC provided by the Pentacam device with a series of different algorithms of machine learning.

SD-OCT parameters and visual field defect in chiasmal compression and the diagnostic value of neural network model

PURPOSE

To evaluate the peripapillary retinal nerve fiber layer (RNFL) and macular ganglion cell-inner plexiform layer (GCIPL) measurements using spectral domain optical coherence tomography (SD-OCT) in patients with chiasmal compression and analyze the diagnostic value of a neural network model.

METHODS

Forty-seven patients with chiasmal compressive disorder were recruited and divided into two groups depending on the visual field defect (perimetric; group 1 and preperimetric; group 2). Fifty-seven normal subjects were also recruited (group 3). Peripapillary RNFL and macular GCIPL were analyzed in each group. A multilayer perceptron was trained using a training dataset and derived a neural network model. The diagnostic performances were compared using the area under the receiver operating curve (AUROC) between each parameters and neural network model.

RESULTS

All macular GCIPL parameters, except inferotemporal GCIPL thickness, were thinner in group 1 than in group 2 and group 3, with barely any difference between group 2 and group 3 parameter values. The diagnostic power of the neural network model, minimum GCIPL, and inferonasal GCIPL were superior when compared with other parameters; the diagnostic values of these three parameters are not significantly different in discriminating the patients and normal control. However, the neural network exhibited the best diagnostic power in distinguishing group 2 and group 3.

CONCLUSION

Macular GCIPL was reduced in chiasmal compression patients with visual field defect which was not evident in the preperimetric state. Neural network model showed superior diagnostic value in discriminating the preperimetric patients from normal control. The results suggest that neural networks may be helpful in the early diagnosis of chiasmal compression.

Objective and subjective diagnostic parameters in the fellow eye of unilateral keratoconus

BACKGROUND

Keratoconus (KC) is usually a bilateral corneal ectatic disease. For significant asymmetric presentation (so called unilateral KC), the fellow eye has the mildest and earliest form of the disease, which is typically called forme fruste keratoconus. The aim of this study was to evaluate the sensitivity and specificity of parameters derived from a Scheimpflug imaging system (Pentacam) as well as the changes in the quality of mesopic vision in the apparently normal fellow eye (forme fruste) to detect the earliest and most sensitive parameters.

METHODS

Patients with clinical keratoconus in one eye and forme fruste keratoconus in the fellow eye were compared to subjects with normal eyes. The patients were examined using a rotating Scheimpflug imaging system (Pentacam).The following parameters were evaluated: keratometry, minimum corneal thickness, pachymetry progression index (PPI), Ambrósio relational thickness (ART), posterior elevation, back difference elevation (BDE) and multimetric D index(D index). Receiver operating characteristic (ROC) curves were analyzed by evaluating the area under the curve (AUC) to detect the sensitivity and specificity of each parameter. Mesopic vision evaluations were performed by contrast sensitivity and glare tests for each group.

RESULTS

A total of 48 patients with clinical keratoconus in one eye and forme fruste keratoconus in the fellow eye and 72normal subjects were evaluated. In the clinical keratoconus eyes, the mean K, back difference elevation (BDE), pachymetric progression index maximum(PPI max), and multimetric D were significantly higher compared to the normal subjects, whereas the corneal pachymetry and Ambrósio relational thickness maximum (ART max) were significantly lower. In the forme fruste eyes, the ROC analysis showed that the AUC values of the mean K, thinnest pachymetry, ARTmax, BDE, D index, and PPI max were 0.82, 0.61, 0.88, 0. 67, and 0.64, respectively. The contrast sensitivity and glare tests were significantly affected in the forme fruste cases.

CONCLUSION

In forme fruste keratoconus eyes, the ART max is considered a highly sensitive objective parameter. Contrast sensitivity and glare is an important subjective test, which is affected in forme fruste patients.

Discriminant Value of Custom Ocular Response Analyzer Waveform Derivatives in Forme Fruste Keratoconus

PURPOSE

To evaluate the performance of corneal hysteresis (CH), corneal resistance factor (CRF), 37 Ocular Response Analyzer (ORA) waveform parameters, and 15 investigator-derived ORA variables in differentiating forme fruste keratoconus (KC) from normal corneas.

DESIGN

Case-control study.

METHODS

Seventy-eight eyes of 78 unaffected patients and 21 topographically normal eyes of 21 forme fruste KC patients with topographically manifest KC in the contralateral eye were matched for age, the thinnest point of the cornea, central corneal thickness, and maximum keratometry. Fifteen candidate variables were derived from exported ORA signals to characterize putative indicators of biomechanical behavior, and 37 waveform parameters were tested. Differences between groups were assessed by the Mann-Whitney test. The area under the receiver operating characteristic curve (AUROC) was used to compare the diagnostic performance.

RESULTS

Ten of 54 parameters reached significant differences between the groups (Mann-Whitney test, P < .05). Neither CRF nor CH differed significantly between the groups. Among the ORA waveform measurements, the best parameters were those related to the area under the first peak, p1area, and p1area1 (AUROC, 0.714 ± 0.064 and 0.721 ± 0.065, respectively). Among the investigator ORA variables, a measure incorporating the pressure-deformation relationship of the entire response cycle performed best (hysteresis loop area, AUROC, 0.694 ± 0.067).

CONCLUSION

Waveform-derived ORA parameters, including a custom measure incorporating the pressure-deformation relationship of the entire response cycle, performed better than traditional CH and CRF parameters in differentiating forme fruste KC from normal corneas.

Epithelial remodeling as basis for machine-based identification of keratoconus

PURPOSE

To develop and evaluate automated computerized algorithms for differentiation of normal and keratoconus corneas based solely on epithelial and stromal thickness data.

METHODS

Maps of the corneal epithelial and stromal thickness were generated from Artemis-1 very high-frequency ultrasound arc-scans of 130 normal and 74 keratoconic subjects diagnosed by combined topography and tomography examination. Keratoconus severity was graded based on anterior curvature, minimum corneal thickness, and refractive error. Computer analysis of maps produced 161 features for one randomly selected eye per subject. Stepwise linear discriminant analysis (LDA) and neural network (NN) analysis were then performed to develop multivariate models based on combinations of selected features to correctly classify cases. The sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were determined for each classifier.

RESULTS

Stepwise LDA resulted in a six-variable model that provided an AUC of 100%, indicative of complete separation of keratoconic from normal corneas. Leave-one-out analysis resulted in 99.2% specificity and 94.6% sensitivity. Neural network analysis using the same six variables resulted in an AUC of 100% for the training set. Test set performance averaged over 10 trials gave a specificity of 99.5 ± 1.5% and sensitivity of 98.9 ± 1.9%. The LDA function values correlated with keratoconus severity grade.

CONCLUSIONS

The results demonstrate that epithelial remodeling in keratoconus represents an independent means for differentiation of normal from advanced keratoconus corneas.

Evaluation of corneal elevation, pachymetry and keratometry in keratoconic eyes with respect to the stage of Amsler-Krumeich classification

AIM

To evaluate corneal elevation, pachymetry and keratometry in keratoconic eyes according to the clinical stage of the disease.

METHODS

This prospective comparative study was performed on one hundred and twenty-six eyes of 83 patients who had keratoconus, and 42 normal eyes of 42 age-matched subjects. Corneal elevation, pachymetry and keratometry were measured using a rotating Scheimpflug camera (Pentacam HR, Oculus) in these eyes. The area under the receiver operating characteristic (AUROC) curves was used to analyse the diagnostic significance of these parameters, with respect to each stage of Amsler-Krumeich classifications. AUROC was calculated to describe the predictive accuracy of the different indices and to determine the cut-off points where sensitivity and specificity were maximised.

RESULTS

Posterior (0.980) and anterior (0.977) elevation differences showed the highest AUROCs, followed by dioptres (D) value (0.941), percentage thickness increase (PTI) 2 mm (0.931), PTI 4 mm (0.927), progression index (0.927), minimal pachymetry (0.923), average keratometry (0.914), anterior elevation (0.909), PTI 6 mm (0.906), posterior elevation (0.898), central pachymetry (0.889), PTI 8 mm (0.870), PTI 10 mm (0.864), corneal thickness spatial profile 2 mm (0.835) and cylinder (0.796). The differences in AUROC curves between anterior and posterior elevation difference measurements and other diagnostic parameters tended to be larger at the earlier stages of keratoconus.

CONCLUSIONS

Anterior and posterior corneal surface height data obtained by enhanced ectasia display, effectively discriminates keratoconus from normal corneas. Elevation difference measurements may provide useful information for improving the diagnostic accuracy of keratoconus, especially in the early stage of the disease.

The Challenges of the Detection of Subclinical Keratoconus at Its Earliest Stage

Undetected subclinical keratoconus (KC) is the main risk factor for iatrogenic ectasia. Many parameters have been proposed to help differentiate normal from subclinical KC corneas. Linear discriminant analysis is a technique that models the difference between different classes of data by looking for linear combinations of variables which best explain the data. The association of surfaces elevation, corneal thickness profile and anterior curvature indices leads to the best sensitivity and specificity for the discrimination between normal and early subclinical KC corneas.

How to cite this article

Gatinel D, Saad A. The Challenges of the Detection of Subclinical Keratoconus at Its Earliest Stage. Int J Keratoco Ectatic Corneal Dis 2012;1(1):36-43.

Corneal biomechanical properties in normal, forme fruste keratoconus, and manifest keratoconus after statistical correction for potentially confounding factors

PURPOSE

To evaluate the difference in corneal biomechanical properties, after controlling for potentially confounding factors, along the spectrum of keratoconic disease as measured by the keratoconus severity score.

METHODS

The corneal biomechanical properties of 73 keratoconic (KCN) eyes of 54 patients, 42 forme fruste keratoconic (FFKCN) eyes of 32 patients, and 115 healthy eyes of 115 age- and sex-matched patients were reviewed retrospectively. The main outcome measures were corneal hysteresis (CH) and corneal resistance factor (CRF).

RESULTS

In the normal group, the mean CH was 11.0 ± 1.4 mm Hg and mean CRF was 11.1 ± 1.6 mm Hg. The FFKCN mean CH was 8.8 ± 1.4 mm Hg and mean CRF was 8.6 ± 1.3 mm Hg. The KCN mean CH was 7.9 ± 1.3 mm Hg and mean CRF was 7.3 ± 1.4 mm Hg. There were statistically significant differences in the mean CH and CRF in the normal group compared with the FFKCN and the KCN groups (P < 0.001) after statistically controlling for differences in central corneal thickness, age, and sex.

CONCLUSIONS

There is a significant difference in the mean CH and CRF between normal and FFKCN corneas after controlling for differences in age, sex, and central corneal thickness. However, there is a significant overlap in the distribution of CH and CRF values among all groups. The biomechanical parameters CH and CRF cannot be used alone but may be a useful clinical adjunct to other diagnostic tools, such as corneal tomography, in distinguishing normal from subclinical keratoconic corneas.

Fluorescein pattern interpretation in keratoconus

PURPOSE

The purpose of this study is to assess the effect of disease severity on how accurately contact lens fluorescein patterns can be interpreted in keratoconus by clinician assessment.

METHODS

Two clinicians evaluated fluorescein patterns on 111 eyes of 60 patients with mild (<45 D, 14 eyes), moderate (45 D to 52 D, 61 eyes,) and severe (>52 D, 36 eyes) keratoconus. The masked clinicians were given six contact lenses in random order, the lens that just cleared the corneal apex (the first definite apical clearance lens), three lenses flatter (in 0.1 mm increments), and two lenses steeper (in 0.1 mm increments) than the first definite apical clearance lens. They ranked the lenses from flattest to steepest, based on the fluorescein patterns. The percentage of lenses correctly ranked was determined using (1) exact match with actual; (2) within 0.1 mm of actual; and (3) within 0.2 mm of actual. Accuracy was assessed as the sum of the squared differences between the actual base curve value and each clinician's ranking. Comparison of the mean percentage correctly ranked and accuracy for each keratoconus severity groups was performed using a mixed linear model.

RESULTS

Neither percentage correctly ranked (using any of the three protocols) nor accuracy was found to be related to severity of keratoconus (p > 0.15 for all comparisons).

CONCLUSIONS

Accuracy of ranking contact lenses in order of base curve radius based on fluorescein pattern assessment by clinicians does not seem to be related to severity of keratoconus. Many factors influence interpretation of fluorescein patterns including all components of the system, fluorescein, tears, cornea, contact lens, external forces, and technique.

Identification of scanning slit-beam topographic parameters important in distinguishing normal from keratoconic corneal morphologic features

PURPOSE

To identify morphologic parameters obtained using scanning slit-beam topography that help distinguish normal from keratoconic corneal morphologic features.

DESIGN

Observational, retrospective, cross-sectional study.

METHODS

This retrospective review examined 207 normal eyes of patients undergoing an initial consultation for primary refractive surgery and 42 eyes with clinical keratoconus (KCN). The following parameters were examined and compared between the two groups: astigmatism, central corneal power, irregularity indices at 3 mm (II3) and 5 mm (II5), maximal posterior elevation (MPE) magnitude and location, thinnest optical pachymetry (TOP) magnitude and location, anterior elevation best-fit sphere (ABFS), posterior elevation best-fit sphere (PBFS), the ratio of ABFS to PBFS, the difference between average inferior and average superior K values at 3 mm and 5 mm in both keratometric (I-S K3 and I-S K5) and tangential (I-S T3 and I-S T5) topographic maps, and skewed radial axis at 3 mm (SRAX3) and 5 mm (SRAX5) of the keratometric topography map.

RESULTS

The II3, II5, MPE magnitude, TOP magnitude, ABFS, PBFS, ABFS-to-PBFS ratio, I-S K at both 3 mm and 5 mm, I-S T at both 3 and 5 mm, and SRAX at 3 mm and 5 mm values were significantly different among the two groups (P < .001). The least-correlated parameters were SRAX3, TOP magnitude, and II3 in the KCN group and I-S K3, amount of astigmatism and MPE magnitude in the normal group.

CONCLUSIONS

Parameters obtained using scanning slit-beam topography may allow improved differentiation of keratoconic from normal corneal shapes, especially when the poorly correlated intragroup parameters are used.

Simulation of Machine-Specific Topographic Indices for Use Across Platforms

PURPOSE

The objective of this project is to simulate the current published topographic indices used for the detection and evaluation of keratoconus to allow their application to maps acquired from multiple topographic machines.

METHODS

A retrospective analysis was performed on 21 eyes of 14 previously diagnosed keratoconus patients from a single practice using a Tomey TMS-1, an Alcon EyeMap, and a Keratron Topographer. Maps that could not be processed or that contained processing errors were excluded from analysis. Topographic indices native to each of the three devices were recorded from each map. Software was written in ANSI standard C to simulate the indices based on the published formulas and/or descriptions to extend the functionality of The Ohio State University Corneal Topography Tool (OSUCTT), a software package designed to accept the input from many corneal topographic devices and provide consistent display and analysis. Twenty indices were simulated. Linear regression analysis was performed between each simulated index and the corresponding native index. A cross-platform comparison using regression analysis was also performed.

RESULTS

All simulated indices were significantly correlated with the corresponding native indices (p < 0.01), with a mean R of 0.84, ranging from 0.42 to 0.99. Cross-platform comparisons were nonsignificant for specific indices and devices.

CONCLUSION

Topographic indices native to three devices were successfully simulated. Cross-platform comparisons may be limited for specific indices.

A New Method for Grading the Severity of Keratoconus

PURPOSE

To define a new method for grading severity of keratoconus, the Keratoconus Severity Score (KSS).

METHODS

A rationale for grading keratoconus severity was developed using common clinical markers plus 2 corneal topographic indices, creating a 0 to 5 severity score. An initial test set of 1012 eyes, including normal eyes, eyes with abnormal corneal and topographic findings but not keratoconus, and eyes with keratoconus having a wide range of severity, was used to determine cutpoints for the KSS. Validation set 1, comprising data from 128 eyes, was assigned a KSS and compared with a clinician's ranking of severity termed the "gold standard" to determine if the scale fairly represented how a clinician would grade disease severity. kappa statistics, sensitivity, and specificity were calculated. A program was developed to automate the determination of the score. This was tested against a manual assignment of KSS in 2121 (validation set 2) eyes from the Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study, as well as normal eyes and abnormal eyes without keratoconus. Ten percent of eyes underwent repeat manual assignment of KSS to determine the variability of manual assignment of a score.

RESULTS

From initial assessments, the KSS used 2 corneal topography indices: average corneal power and root mean square (RMS) error for higher-order Zernike terms derived from the first corneal surface wavefront. Clinical signs including Vogt striae, Fleischer rings, and corneal scarring were also included. Last, a manual interpretation of the map pattern was included. Validation set 1 yielded a kappa statistic of 0.904, with sensitivities ranging from 0.64 to 1.00 and specificities ranging from 0.93 to 0.98. The sensitivity and specificity for determining nonkeratoconus from keratoconus were both 1.00. Validation set 2 showed kappa statistics of 0.94 and 0.95 for right and left eyes, respectively. Test-retest analysis yielded kappa statistics of 0.84 and 0.83 for right and left eyes, respectively.

CONCLUSION

A simple and reliable grading system for keratoconus was developed that can be largely automated. Such a grading scheme could be useful in genetic studies for a complex trait such as keratoconus requiring a quantitative measure of disease presence and severity.

Automated Decision Tree Classification of Corneal Shape

PURPOSE

The volume and complexity of data produced during videokeratography examinations present a challenge of interpretation. As a consequence, results are often analyzed qualitatively by subjective pattern recognition or reduced to comparisons of summary indices. We describe the application of decision tree induction, an automated machine learning classification method, to discriminate between normal and keratoconic corneal shapes in an objective and quantitative way. We then compared this method with other known classification methods.

METHODS

The corneal surface was modeled with a seventh-order Zernike polynomial for 132 normal eyes of 92 subjects and 112 eyes of 71 subjects diagnosed with keratoconus. A decision tree classifier was induced using the C4.5 algorithm, and its classification performance was compared with the modified Rabinowitz-McDonnell index, Schwiegerling's Z3 index (Z3), Keratoconus Prediction Index (KPI), KISA%, and Cone Location and Magnitude Index using recommended classification thresholds for each method. We also evaluated the area under the receiver operator characteristic (ROC) curve for each classification method.

RESULTS

Our decision tree classifier performed equal to or better than the other classifiers tested: accuracy was 92% and the area under the ROC curve was 0.97. Our decision tree classifier reduced the information needed to distinguish between normal and keratoconus eyes using four of 36 Zernike polynomial coefficients. The four surface features selected as classification attributes by the decision tree method were inferior elevation, greater sagittal depth, oblique toricity, and trefoil.

CONCLUSION

Automated decision tree classification of corneal shape through Zernike polynomials is an accurate quantitative method of classification that is interpretable and can be generated from any instrument platform capable of raw elevation data output. This method of pattern classification is extendable to other classification problems.

Repeatability of Corneal Topography Measurement in Keratoconus with the TMS-1

PURPOSE

The purpose of this study was to report the test-retest variability of simulated indices derived from the TMS-1 topography instrument (Tomey Technology, Waltham, MA) in keratoconus subjects enrolled in the Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study.

METHODS

Four images were taken at an initial visit and at a repeat visit several weeks later. From these images, 17 indices were simulated from published formulas. Mixed-model analysis was used on test-retest data from the TMS-1 videokeratography instrument during the baseline year. This analysis yields estimates of within- and between-visit variability.

RESULTS

Repeatability analysis revealed that within-visit standard errors were 1.0 to 5.9 times greater in keratoconus eyes than in normal controls when two images were analyzed from each visit. These values changed only slightly when more images were used. The ratio of between-visit standard errors of the indices were nearly equally greater than normal controls for (0.9-4.6 and 0.9-4.3) two images per eye and all images per eye, respectively.

CONCLUSIONS

These results suggest that the repeatability of simulated indices derived from TMS-1 topography in keratoconus subjects is poorer than in normal controls.