Often when a new or growing technology is released we are confronted with questions; traditional or modern, book or computer, digital or analogue? In part these questions result from fear, fear of the new and fear of losing the old, the known. However, in reality, these questions are a milestone, a marker of a technology which has not yet been fully incorporated into practice. One day these technologies will cease to be a question and simply transition to a method, as their predecessors did before them. Computational analysis is such a technology. With the ever enhancing capabilities of computers, followed quickly by mechanical limitations, we are asked to assess the value of two separate methodologies, the known and the new, computational analysis or curatorial expertise. I reject this choice.
Despite the rapid advance in technology and its increasing innovation there is no doubt that a computer cannot do everything a human can do. When it comes to curating images the human eye sees an image and categorises it based on a number of influential factors such as culture, social conditioning and experience. However, humans can also recognise a multiplicity of term, that one word can be synonymous or interchangeable with another is a given of human language. Therefore, for argument’s sake sometimes curatorial expertise is necessary in order to generate a comprehensive catalogue.
However, just as computers can’t do everything a human can do, humans can’t do everything a computer can do. For example, in Oxford Dr Christoffer Nellaker and his team are developing computer vision algorithms which will analyse photographs of faces for disease-relevant phenotypes (Oxford University Innovation). Categorised under the field of eHealth Dr Nellaker’s project will use these algorithms to help medical professionals make easier and more accurate diagnoses of rare genetic diseases. The programme uses machine learning to create a multidimensional space shaped to account for deceptive variations such as lighting, pose, occlusions, and image quality. Similarly, computational imaging techniques have been applied to patients in order to provide automated decision support for experts when trying to grade the severity of pathologies such as diabetic retinopathy and macular degeneration (Dessauer & Dua 39). Such technologies allow these experts to create pattern recognition which enhance their ability to make informed case decisions based on common characteristics and previous cases in a much more efficient way.
Therefore, both curatorial expertise and computational analysis have their benefits. Just as a computer cannot interpret the various interpretations associated with one image, an optometrist cannot perform fifty eye exams and accurately diagnose the severity of a pathology. This is why I don’t believe it is a choice between one or the other, it is a compromise, it is a discussion but most of all it revolves around implementation. One method does not supercede the other, both have their merits and both have their pitfalls. The real question is how do we or can ever perfectly combine the two?
Dessauer, Michael & Dua, Sumeet, ‘Computational Methods for Feature Detection in Optical Images.’ Computational Analysis of the Human Eye with Applications. Ed. Duan Semeet et al. Singapore: World Scientific Publishing, 2011. 39 – 88.
Nellaker, Christoffer, ‘Diagnosis of rare diseases with computational analysis of photographs.’ Oxford University Innovation (http://innovation.ox.ac.uk/wp-content/uploads/2015/04/Aiding-Diagnosis-of-Rare-Diseases-Through-Computational-Analysis-of-Photographs-Poster.pdf) Web. 09 December 2016.