OCR Camera System

ARUP has invented an automated camera system that uses sophisticated software and optical character recognition (OCR) technology to identify specimens that are potentially mislabeled for the patient name. A prototype system was operated at ARUP on our old track system from October, 2012 to October 2014. During that time more than 3 million specimens were photographed and the images analyzed by the OCR system. Click here to link to a separate section of the webpage that describes the prototype system and the results of the two year trial period. A video clip of that system can also be viewed by clicking here.

Four new OCR systems are now being built to operate on the new MagneMotion Automated Transport and Sorting System. The anticipated date for installation is later in 2015 or early 2016. The operational plan is that each tube on the track will pass through one of these OCR systems before going to a sorter or thawing and mixing work cell. Most images are “passed” by the OCR analysis meaning there is no discrepancy between the patient name on the customer’s label and the patient name in the ARUP LIS. However, if an image is not passed by the OCR system (this is called a “fail”), inspection of the failed image by a trained employee is required. This inspection occurs soon after the image is obtained and before any testing would be performed. If the inspection of the image shows that the specimen was mislabeled (there is a discrepancy between the patient name on the customer’s label and the patient name in the ARUP LIS), that specimen will be retrieved and relabeled prior to testing.

There are many reasons why an image can fail OCR analysis even though the specimen is correctly labeled. Only one in approximately 2,000 images that fail OCR analysis is actually from a mislabeled specimen. The vast majority of OCR failures are caused by issues with client labels. These reasons include:

  • Non-standard fonts, small fonts, and fonts with serifs
  • Poor quality labels
  • Handwriting or marks on the label that touch the patient name
  • Letters and characters too close to each other, horizontally and vertically, due to clients attempting to crowd as much information on the label as they can
  • Name truncations
  • Colored labels, or labels with colored stripes
  • Splitting the patient name on two lines of the label
  • Wrapping the patient name at 90 degree to the length of the tube
  • Handwritten labels

It should be noted that these issues that prevent the OCR system from passing correctly labeled specimens, are often the same issues that contribute to mislabeled specimens in all laboratories, potentially causing harm to patients. Please click here to read ARUP’s recommendations on how clients can improve the quality of labels in their own laboratory including the labels placed on specimens sent to ARUP.