Ursa Sapiens
  • The Blog
  • Team
  • About Us
  • Categories
  • Blog Archives
    • December 2014
    • November 2014
    • April 2014
    • March 2014
    • February 2014
    • January 2014
    • December 2013
  • Brown TTH

Putting Your Life in Computerized Hands: A.I. for Intraoperative Tumor Diagnosis

3/28/2020

0 Comments

 
Written by: Devin Juros
Edited by: Sisasenkosi Mandi
Picture
As anyone who has been through major surgery can attest, putting your life in the hands of someone else, even when that person is a highly skilled surgeon attempting to save your life, is difficult. How, then, would people react to putting their life in the inanimate hands of a computer? As artificial intelligence (AI) is developed and improved, new uses for the technology are being established, including in surgery. Recently, T Hollon and others developed effective technology for intraoperative brain tumor diagnosis using laser imaging and AI.
During surgery, neurosurgeons often take biopsies of removed tumors for preliminary tumor diagnosis. These intraoperative diagnoses aid neurosurgeons in determining the type of cancer present and whether they should continue removing brain matter around the affected area or not during surgery. The traditional method of analyzing these biopsies involves a neuropathologist taking 30 minutes to freeze, stain, prepare, and assess the specimen. If several biopsies need to be taken during a surgery, then the patient has to stay under anesthesia for longer, which can be dangerous. Each extra hour under general anesthesia increases the risk of complications (like pneumonia, strokes, or heart attacks) by 18-36% [5]. 

T Hollon’s novel method takes biopsy images with a laser imaging technique and then analyzes these images with AI, which can differentiate the types of tissue in the biopsy, including the type of cancer present. The total processing time: one to two minutes. In a trial study, this method had a similar correct diagnosis rate to neuropathologists (94.6% to 93.9%) [4]. With this new technique, neurosurgeons can make more numerous intraoperative diagnoses to ensure they are removing tumors more completely and patients can stay under anesthesia for a shorter time. This new method could greatly improve neurosurgical outcomes, but would you put your life in the hands of this computer?

One’s instinctive reaction to the facts may be yes; if artificial intelligence can do the job of a neuropathologist just as well and reduce time under anesthesia, why not? However, upon further consideration, one will likely gain an eerie feeling about allowing an inanimate object to control one’s fate. 

Many people have a similar eerie feeling when considering the potential of self-driving cars. Self-driving cars have the potential to revolutionize driving and decrease the six million annual car crashes in the United States [2]. 94% of all of these motor vehicle accidents are caused by human error [1]. Driverless cars have the potential to virtually eliminate these human error accidents through precise algorithms. In addition, driverless cars remove drunk driving, road rage, and tired driving from the death toll. Yet, besides these glowing potential benefits, 42% of people still say they would never ride in a driverless car [6]. We are faced once more with a paradox: why does something that prevents injuries and deaths cause such widespread doubt? Again, the human aspect is missing.

In a life-threatening situation like surgery or a near-accident, we want a human at the helm, even if the computer performs better statistically. Psychological studies have shown that we trust people more if they have a rule-based morality that considers the morality of the action and not just the consequences when making decisions. However, AI is strictly consequentialist, considering only the outcomes in its calculations, making it less trustworthy on a moral level [3]. Perhaps, quite humanly, we also want a human to blame if things go wrong. How could you pin your suffering on a lifeless computer? 

Or, people may be worried about AI permeating too far through society. As artificial intelligence is used for an increasingly diverse list of tasks, some fear that humans are losing their place in society. One day, all cars on the road may be driverless. Or, all surgeries may be diagnosed and even performed by AI and machines. Where will artificial intelligence stop?

Yet, are these fears irrational? If AI for intraoperative tumor diagnosis improves people’s well-being, is it not our duty to support it? At a certain point, it may be necessary to put our reservations aside in order to improve surgical outcomes.
​

In the meantime, while some people still hold viscerally pessimistic reactions to artificial intelligence in areas like surgery, it may be beneficial to combine humans and artificial intelligence. As artificial intelligence is still being improved, this combinatorial approach of balancing the mass processing power and precision of artificial intelligence with the reasoning and oversight of humans may lead to the best outcomes and mitigate the negative instinctive reactions of the public. While advances continue to be made in the field of surgery using AI, it may be a long time before technologies such as the one created by T Hollon are widely available and applied routinely in surgeries.

Works Cited: 

[1] Brown, B. (2017, October 7). 2016 NHTSA Fatality Report Adds to Evidence for Self-driving Cars. Retrieved February 14, 2020, from https://www.digitaltrends.com/cars/2016-nhtsa-fatality-report/

[2] Car Accident Statistics in the U.S.: Driver Knowledge. (n.d.). Retrieved February 14, 2020, from https://www.driverknowledge.com/car-accident-statistics/

[3] Crockett, M. (2017, April 24). Why are we reluctant to trust robots? Retrieved February 23, 2020, from https://www.theguardian.com/science/head-quarters/2017/apr/24/why-         are-we-reluctant-to-trust-robots

[4] Hollon, T. C., Pandian, B., Adapa, A. R., Urias, E., Save, A. V., Khalsa, S. S. S., … Orringer, D. A. (2020, January 6). Near real-time intraoperative brain tumor diagnosis using         stimulated Raman histology and deep neural networks. Retrieved February 13, 2020,         from https://www.nature.com/articles/s41591-019-0715-9

[5] Morris, A. (2020, January 16). Are self-driving cars safe? Expert on how we will drive in the future. Retrieved February 14, 2020, from https://theconversation.com/are-self-driving-cars-safe-expert-on-how-we-will-drive-in-the-future-128644

[6] Sharoky, C. (2005, July 20). Time spent under anesthesia could up risk. Retrieved February 14, 2020, from https://www.upi.com/Health_News/2005/07/20/Time-spent-under-anesthesia-could-up-risk/27551121885501/

[7] Image source: https://commons.wikimedia.org/wiki/File:Artificial_Neural_Network_with_Chip.jpg ​

0 Comments



Leave a Reply.

Powered by Create your own unique website with customizable templates.
  • The Blog
  • Team
  • About Us
  • Categories
  • Blog Archives
    • December 2014
    • November 2014
    • April 2014
    • March 2014
    • February 2014
    • January 2014
    • December 2013
  • Brown TTH