Royal Philips ($PHG) and Hitachi Data Systems will collaborate to develop a speedy, universal medical image management system that it expects will be on the market during the second half of this year.
Images account for a huge portion of healthcare data--and players are placing big bets to be able to store, organize and make them accessible. But the real prize will go to whoever can deliver on the promise of deep learning and artificial intelligence to glean useful patient and population information from this ever-growing mountain of healthcare data to actually improve tools for healthcare providers.
Like Philips, IBM Watson Health ($IBM) made a major commitment to medical image management with its $1 billion acquisition of Merge Healthcare last August. The plan is to use the machine learning capabilities of its Watson to add value to all that unstructured data.
 |
| Jeroen Tas, CEO of Healthcare Informatics Solutions and Services at Philips |
Philips is quite clear that its system will be a universal one that's interoperable, regardless of vendor, across systems and imaging devices as well as across various departments within healthcare organizations. And it must be fast--enabling physicians to see relevant medical images in less than three seconds.
"There is increasingly a need from our customers to have a vendor neutral solution," Jeroen Tas, CEO of Healthcare Informatics Solutions and Services at Philips, told FierceMedicalDevices in an interview. "Ultimately, the customer wants an integrated view of the patient; they want to make sure devices interoperate. It's not all Philips, GE or Medtronic devices. We are all part of the big picture so it's necessary to interoperate. The same applies to the EMR players. It's in the best interest of patients and healthcare providers and, ultimately, in our interest to create richer and better solutions."
The eventual goal is to offer an integrated view of a patient's medical condition, aided by deep learning and artificial intelligence, to improve patient outcomes.
"I don't think we're replacing radiologists any time soon. But we are applying it today," said Tas. "You move a cursor over an image and it will identify the liver and you can click on liver inspector, create a quantification and if there's a cancer identify and quantify it. We have a product that launched a couple of months ago that can identify whether cancer cells are dead or living--which is important to see if chemotherapy is really working. We are increasingly applying deep learning to create stronger contexts, to help with interpretation and to allow better assessment of what's normal and identify slight deterioration that may be hard to see with the human eye."
Philips manages more than 135 billion medical images. And it says that many of its health systems are generating more than 2 million medical images per week. These include X-ray, CT and MRI scans made by radiology, cardiology and oncology departments as well as digital photos taken by dermatology and plastic surgery during a consultation.
But image management and smarter analytics are just the start. Next, healthcare providers will also need to combine and comb through all the massive quantities of data generated by genomics.
"When you combine what we are doing with imaging with digital genomics, we can get up to even a terabyte per patient for very complex genomics. And then when you start using deep learning on those images, you start aggregating 100,000 of those one terabytes," concluded Tas.
- here is the announcement