Missing relevant prior studies account for around 10 percent of all lawsuits related to medical imaging.

PACS Vendor IMAGE Information Systems is setting out to eliminate this challenge with the release of its iQ-VIEW PRO 3.1, a reading software imbued with AI capabilities for sussing out and automatically detecting prior studies of patients in PACS environments.

"The subtle imaging changes are the challenging ones, so image comparison with prior studies is important for medical image analysis," Dr. Arpad Bischof, a radiologist and a director of IMAGE Information Systems, told HCB News. "If I read a CT scan of the heart, and there was a prior CT scan of the chest, or an x-ray of the thorax, the system will highlight both prior studies as relevant, as the system 'knows' that the heart is a part of both the chest and the thorax, and highly relevant for comparison. This helps save 15-30 minutes per day per radiologist and prevents priors from being missed, leading to better reading."
Identifying relevant prior studies can be complex, such as in the case of a thorax CT scan, which may require access to a chest X-ray, a whole body PET/CT scan, a cardiac MR or more exams. Imaging device vendors also refer to body parts with different names from one another, creating more confusion in PACS systems.

To top off the matter, a recent study of a German University Hospitals PACS found that 80 percent of elderly patients had non-unique patient identifiers in the system, while 12 percent had two duplicate identities and six percent had three to seven identities. Multiple forms of identity are attributed to typing errors, change in last names, and several middle names.

The software is a component of the iQ-System PACS, trained to identify by default which anatomical parts connect to one another throughout the entire human body, such as the finger to the metacarpal of the hand and the hand to the arm, referred to as the upper limb or upper extremity.

This AI capability enables the platform to automatically identify relevant prior studies of a patient in 1-2 seconds in PACS with up to more than one million studies.

It also can recognize synonyms of body parts including their translation in foreign languages, and auto-detects spelling of patient names in other languages, comparing multiple versions of names to dates of birth to verify if the patient is the same despite the presence of different patient IDs.

In the case of a patient named Sarah Newman, for instance, the software can identify her whether she has multiple IDs with variations in her name such as Newmann, Sandra versus Newman, Sandra. It can also take into account names with foreign characters such as with Ruß, Sandra versus Rusz, Sandra.

"Even if the patient comes using one patient ID to hospital A and using another patient ID to hospital B and a similar writing, iQ-VIEW PRO 3.1 will do proper matching with an accuracy of 1:30.000 to 1:100.000, which is far better and faster than the accuracy of a human," said Bischof. "We analysed, during development of this product, two million data sets from 800 imaging centers around the world in order to reach this accuracy level."

iQ View PRO is compatible as a diagnostic multi-modality reading station with any DICOM-compliant PACS. iQ-VIEW is equipped with more than 20 user interface languages and can be purchased in 112 countries from a network of factory trained service partners, or directly from IMAGE.

iQ-VIEW PRO 3.1 comes with 25 percent faster query/retrieve, faster zoom/pan, bug fixes, and several smaller advantages. It is CE-marked and FDA-cleared as part of the iQ-System PACS.

Health IT Homepage