Computer-aided measurement of psoriatic lesion area in a multicenter clinical trial—Comparison to physician's estimations
Introduction
Psoriasis is a common immune-mediated chronic skin disease that primarily affects knees, elbows, scalp, hands, feet and lower back. As psoriasis is a complex disease, there is no adequate animal model to test newly emerging antipsoriatic treatments [1]. Recently the animal models were developed which enable the study of etiology of psoriasis [2] but not the study of therapeutic effects. In clinical trials, treatment efficacy is determined by assessing signs such as the area of psoriatic lesion and intensity of erythema, infiltration/plaque thickness, and desquamation. These signs are not easy to measure and are generally estimated by physicians. There are three main sources of variability in clinical trials, which influence the number of patients that must be involved: variability between patients, between-days variability and variability of estimation. Between-days variability is caused by external factors which may exacerbate psoriasis [3]. This can partially be eliminated by calculating the average of multiple repeated estimations of signs at different days at the end or during the treatment. Variability (error) of sign estimation can also be minimized by repeating the estimation on several days or by several physicians, but this complicates the organization of the clinical trial. Previous studies have indicated wide variations between observer's estimates of involved surface area [4], [5], [6], [7], [8]. Computer-aided image analysis (CIA) was used by several authors to minimize the variability of psoriatic area estimation [6], [7], [8], [9], [10], [11], [12], [13], [14]. Both photographs and drawing were used in CIA. The methods for measurement of the intensity of other psoriasis signs were also developed [9], [11], [14]. Curved body sites and consequently the uneven illumination of photographs (shadows) is the main obstacle for efficient CIA. Distortion due to the curved body sites may be corrected by conversion factor obtained from reference points marked on the skin [11] however previous study show that this error may be neglected [7]. The problem of uneven illumination was minimized by green filter [11] or preferably by color segmentation techniques [12]. In the preliminary study we found that irregularities on the skin (such as wrinkles and scars) introduce false positive result in CIA, whereas the scales in the psoriatic lesions introduce false negative result. Most of CIA methods are expensive, time consuming and thus unfeasible in large-scale clinical trials [12], [15]. In the present study we optimized a method of psoriatic area measurement by combining (1) the physician's proficiency to accurately determine the edge of the lesion regardless of irreproducible light conditions and (2) the ability of image analysis to accurately measure a defined area. Although the proposed method requires digital cameras and computerized off-line analysis it is feasible in large-scale multicenter clinical trials.
Psoriasis area and severity index (PASI), which combines the psoriatic area with three psoriatic signs has been introduced [16] to make a psoriasis grading more accurate. This index is the most frequently used measure for evaluating psoriasis and its therapy. We applied an adapted PASI index, where the psoriatic area was not converted into an area grade, but was maintained as a continuous variable, similar to modification suggested by Jacobson and Kimball [17]. This is to prevent the loss of data in the process of transferring an area into the area score.
Section snippets
Patients and a clinical trial
A total of 46 consecutive patients (28 males and 18 females, age 18–79 years, average 51.8 years) with chronic psoriasis vulgaris entered this randomized double-blind, vehicle-controlled, parallel group study from January 2004 to March 2004 in four study centers. The patients had psoriasis of one or more of the following regions: upper extremities, lower extremities or trunk. The PASI index was between 5.2 and 17.8 (average 9.7). They received no treatment with topical antipsoriatic therapy
Results and discussion
There are at least three theoretical advantages of the area measured on pictures (ComPA) over conventional psoriatic area estimated by a physician (PA). (1) The edge of psoriatic area cannot be objectively determined. This is particularly difficult in case of spotty lesions and different physicians can determine the edge of lesion differently. In ComPA approach, all the patients from four clinics were evaluated on photographs by one physician. (2) Although the trial was double-blind, the
Acknowledgements
The authors wish to thank to Lek Pharmaceuticals d.d., who sponsored the trial. We are grateful to Valerija Balkovec, MD; Ana Benedičič Pilih, MD; Maja Gabrič –Zirkelbach, MD; Assist. Prof. Tomaž Lunder, MD, PhD; Jovan Miljković, MD; Božana Podrumac, MD; Ranka Popovič, MD and Ida Prelog, MD for contribution of their patient's data.
References (17)
- et al.
20 MHz sonography, colorimetry and image-analysis in the evaluation of psoriasis-vulgaris
J Dermatol Sci
(1995) - et al.
A comparison of subjective and objective measures of reduction of psoriasis with the use of ultrasound, reflectance colorimetry, computerized video image analysis, and nitric oxide production
J Am Acad Dermatol
(1997) - et al.
Psoriasis: the future
Br J Dermatol
(2001) - et al.
Spontaneous development of psoriasis in a new animal model shows an essential role for resident T cells and tumor necrosis factor-alpha
J Exp Med
(2004) - Swerlick RA, Lawley TJ. Eczema, Psoriasis, Cutaneous Infections, Acne, and Other Common Skin Disorders (Chapter 56) in:...
- et al.
Assessment of disease progress in psoriasis
Arch Dermatol.
(1989) - et al.
Measurement of body surface area involvement in atopic eczema: an impossible task?
Br J Dermatol
(1999) - et al.
Assessment of area of involvement in skin-disease—a study using schematic figure outlines
Br J Dermatol
(1993)
Cited by (35)
Current and Emerging Imaging Techniques for Neurofibromatosis Type 1–Associated Cutaneous Neurofibromas
2023, Journal of Investigative DermatologyPsLSNetV2: End to end deep learning system for measurement of area score of psoriasis regions in color images
2023, Biomedical Signal Processing and ControlValidation of artificial intelligence application in clinical dermatology
2022, Journal of the American Academy of DermatologyAutomated psoriasis lesion segmentation from unconstrained environment using residual U-Net with transfer learning
2021, Computer Methods and Programs in BiomedicineJoint AAD–NPF Guidelines of care for the management and treatment of psoriasis with topical therapy and alternative medicine modalities for psoriasis severity measures
2021, Journal of the American Academy of DermatologyCitation Excerpt :Recommendations on the use of BSA and its supporting evidence can be found in Tables XXVIII and XXIX. BSA can be overestimated, particularly by untrained providers.219-223 Nevertheless, BSA assessment has good intra-rater reliability.214,224
A cascaded deep convolution neural network based CADx system for psoriasis lesion segmentation and severity assessment
2020, Applied Soft Computing JournalCitation Excerpt :In current practice, dermatologist utilizes the subjective measurement which requires visual and haptic cues for psoriasis lesion segmentation and it is severity measurement. However, these methods are time-consuming and might provide unreliable results due to its highly subjective nature [5–9]. Thus, objective methods have been proposed by various researchers in the recent past.