MelaFind, a device designed for use by dermatologists for in-vivo detection of suspicious pigmented skin lesions, has been narrowly approved by an FDA advisory panel. Here is an excerpt from the article (see: Panel Splits over Skin Cancer Detection Device):
An FDA advisory panel has voted 8-7, with one member abstaining, to recommend approval for an experimental skin cancer detection system called MelaFind. A vote so close generally is not considered an endorsement in the eyes of FDA officials, who have the final say in whether to approve the device. The FDA does not have to follow the advice of its advisory committees, but it often does. Some panelists on the [committee] were concerned that relying on a device to detect melanoma could lead to unnecessary biopsies, or worse: missed skin cancers. Those who voted against the device also were concerned that non-dermatologist physicians might start using it. "I have a problem with this," said panelist Lynn Drake, MD, a dermatologist at Massachusetts General Hospital and a lecturer at Harvard, who voted against approval. "This disease kills people. I think we need to be very careful about approving something that might replace a human's judgment." But other panelists felt that even though the device proved only slightly superior at detecting melanoma in suspicious lesions than a physician alone, having another tool to aid in skin cancer detection would ultimately benefit patients....The noninvasive device uses a dermascope with near-infrared light to image the skin through a thin layer of alcohol or oil while a digital camera captures images of the lesion. A computer connected to the dermascope displays the image on a screen in the physician's office. Software sorts out various patterns and analyzes lesion color, shape, and consistency to assign a risk that a lesion is a melanoma. MELA Sciences, the device's manufacturer, says MelaFind can visualize lesion structures under the skin.
This news item interested me for a few reasons. First of all, MetaFind is an example of in-vivo imaging technology that, at some time in the future, may replace the pathologist and histopathology in the diagnosis of various types of lesions. Other examples of such technology include molecular imaging (see: The Role of Molecular Imaging Versus Histopathology in Diagnosis) and in-vivo endoscopic imaging. Neither of these two technologies will replace the pathologist soon but it's not hard to see a trend here.
Two of the objections to the product by the panel members, quoted in the article, also caught my attention. One is the "guild defense" argument cited by a dermatologist that non-dermatologists might start using the device and thus take away work from specialists. In my opinion, this can't happen fast enough. I believe that much of the initial triage of patients during office visits should be managed by specially trained nurses. It would be an advantage for all of us if initial skin-checks could be accomplished by such nurses using a product like MelaFind. A dermatologist can be brought into the room, when necessary, at the end of the visit to confirm the clinical findings. This would be a very efficient way to address the current epidemic of melanomas.
Secondly, I was struck by the objection of one of the panel members that a downside of the device was that technology will "replace human judgement." In my opinion, only poorly-designed technology poses such a risk. Well-designed technology should always improve human performance but will provide an option to override the technology if and when human judgement is superior. This is the case when using computer algorithms to quantify tissue biomarkers in digital pathology. However and in such instances, the human operator needs to have a strong rationale for the override. After all, human expertise is used to develop the computer algorithms and the intelligence of the current operator may be superior to that which was used to develop the algorithms (see: In-Vitro Biomarkers vs. In-Situ Biomarkers). Moreover, the algorithms may also not be working properly or inadequately designed.
The idea that pathologist can be replaced by this is not a good idea. Men cannot be replaced by machines. That is one thing that must be put into our minds.
Posted by: Cecilia Whitney | January 19, 2013 at 08:15 PM
It is a useful device to lessen the work of Dermatologist.
Posted by: Patricia Allen | December 08, 2012 at 02:09 AM
Melanomas are very difficulte to diagnose histologically because no criterium is applicable to all lesions. These people will find the same thing with their fancy device.
Posted by: EM | November 25, 2010 at 04:35 PM
You say that human judgement can be used to override the software, but we are fast approaching the point (indeed, we may even have already arrived)at which the data is so complex that only a computer can handle it. Therefore we will have to start using computers/artificial intelligence to design new algorithms and interpret data. At that point human judgement will not be able to override the computer since humans will not understand the complexities of the algorithm/software. Just think how difficult its becoming to debug new soft ware that has millions of line s of code. Correct me if I'm wrong, but to my knowledge we cannot use gene arrays without computers to interpret the data. So just like pathologists might be obsolete in some areas (for example the use of reflectance technology in GI endoscopes that can "see" into tissues may mean that gastroenterologists will be interpreting their own virtual biopsies)human judgement may (will) become obsolete too. It's just a matter of time. Human operators will only be button pushers. Not pessimistic, just realistic.
Posted by: Ajit Alles | November 23, 2010 at 10:08 AM
A new free online CME by Johns Hopkins University School of Medicine is being directed towards pathologists who have patients affected by non-small-cell lung cancer (NSCLC). The online activity is free and will consist of 4 different activities.
Each activity will present different case scenarios using the latest treatment options and diagnostic markers to help pathologists improve patient outcomes.
To learn more about this free online CME activity from Johns Hopkins University School of Medicine, and how you as a pathologist can get up-to-date information on the diagnosis and management of patients with NSCLC please visit: http://www.jhasim1.com/2010-14/index1.aspx .
Posted by: Hollyanne | November 19, 2010 at 12:43 PM