What is the role of automated instruments in melanoma diagnosis?
An automated diagnostic instrument is defined as one that requires minimal or no input from the clinician to achieve a diagnosis. Each automated instrument offers different technology with differing diagnostic ability. Guidelines for assessing such instruments have been published. To date, only 2 studies have been reported comparing clinician diagnosis or management with machine diagnosis with an adequate sample size to assess both specificity and sensitivity for the diagnosis of melanoma.
Summary of systematic review results
The MelaFindTM system, a digital multispectral image analysis device for the use on suspicious pigmented melanocytic lesions, was directly compared to specialists’ diagnosis in a prospective multicentre clinical trial. Here, lesions were recruited (analysed) if they were scheduled for biopsy, usually because of clinician concern. The measured sensitivity of MelaFindTM was 98.4% (125 of 127 melanomas; 95% lower confidence bound 95.6%) which achieved the pre-trial primary aim and had a superior specificity (9.9%) to clinicians’ (3.7%); p=0.02.
The NevisenseTM system, an electrical impedance device for the use on lesions, irrespective of pigmentation, where a diagnosis of melanoma needs exclusion, underwent a prospective multicentre clinical trial in a specialist setting. The observed sensitivity of NevisenseTM was 96.6% (256 of 265 melanomas; 95% lower confidence bound 94.2%) with an observed specificity of 34.4%. Again, lesions were recruited if they were scheduled for biopsy, but a direct comparison with the recruiting clinician’s diagnosis was not performed.
In both of the above systems high false positive rates with the highly prevalent seborrhoeic keratoses may cause a significantly poorer specificity when used by non-experts in the field. This has yet to be investigated. Indeed, currently there is no data on the use of these instruments in clinical trials in a primary care setting.
The effect of adding the MoleMateTM system, a digital image analysis device, to suspicious pigmented lesions in primary care, was assessed in a multicentre randomised clinical trial. The primary endpoint was the effect of the device on the proportion of appropriately referred lesions, where the secondary care experts decided to biopsy or monitor, which did not differ significantly between those lesions being measured by the device (56.8% 130/229) or not (64.5% 111/172); p=0.12. The proportion of benign lesions appropriately managed and the percentage agreement with an expert decision to biopsy or monitor also did not significantly differ between use and non-use of the device. 18/18 melanomas were appropriately referred in the intervention group and 17/18 in the control group.
Evidence summary and recommendations
|To date, only 2 studies have been reported comparing specialist clinician diagnosis with an automated machine diagnosis with an adeqaute sample size to assess both specificity and sensitivity for the diagnosis of melanoma.||II||, |
- Rosado B, Menzies S, Harbauer A, Pehamberger H, Wolff K, Binder M, et al. Accuracy of computer diagnosis of melanoma: a quantitative meta-analysis. Arch Dermatol 2003 Mar;139(3):361-7; discussion 366 Available from: http://www.ncbi.nlm.nih.gov/pubmed/12622631.
- March J, Hand M, Grossman D. Practical application of new technologies for melanoma diagnosis: Part I. Noninvasive approaches. J Am Acad Dermatol 2015 Jun;72(6):929-41; quiz 941-2 Available from: http://www.ncbi.nlm.nih.gov/pubmed/25980998.
- Monheit G, Cognetta AB, Ferris L, Rabinovitz H, Gross K, Martini M, et al. The performance of MelaFind: a prospective multicenter study. Arch Dermatol 2011 Feb;147(2):188-94 Available from: http://www.ncbi.nlm.nih.gov/pubmed/20956633.
- Malvehy J, Hauschild A, Curiel-Lewandrowski C, Mohr P, Hofmann-Wellenhof R, Motley R, et al. Clinical performance of the Nevisense system in cutaneous melanoma detection: an international, multicentre, prospective and blinded clinical trial on efficacy and safety. Br J Dermatol 2014 Nov;171(5):1099-107 Available from: http://www.ncbi.nlm.nih.gov/pubmed/24841846.
- Walter FM, Morris HC, Humphrys E, Hall PN, Prevost AT, Burrows N, et al. Effect of adding a diagnostic aid to best practice to manage suspicious pigmented lesions in primary care: randomised controlled trial. BMJ 2012 Jul 4;345:e4110 Available from: http://www.ncbi.nlm.nih.gov/pubmed/22763392.