WEBVTT

00:00:01.333 --> 00:00:05.150
Computer algorithms today
are performing incredible tasks

00:00:05.174 --> 00:00:09.682
with high accuracies, at a massive scale,
using human-like intelligence.

00:00:10.053 --> 00:00:13.863
And this intelligence of computers
is often referred to as AI

00:00:13.887 --> 00:00:15.736
or artificial intelligence.

00:00:15.760 --> 00:00:20.427
AI is poised to make an incredible impact
on our lives in the future.

00:00:20.911 --> 00:00:24.807
Today, however, we still
face massive challenges

00:00:24.831 --> 00:00:28.325
in detecting and diagnosing
several life-threatening illnesses,

00:00:28.349 --> 00:00:30.854
such as infectious diseases and cancer.

00:00:32.023 --> 00:00:34.277
Thousands of patients, every year,

00:00:34.301 --> 00:00:37.434
lose their lives due to
liver and oral cancer.

00:00:37.894 --> 00:00:40.577
Our best way to help these patients

00:00:40.601 --> 00:00:44.910
is to perform early detection
and diagnoses of these diseases.

00:00:45.871 --> 00:00:50.260
So how do we detect these diseases today
and can artificial intelligence help?

00:00:51.966 --> 00:00:55.592
In patients who, unfortunately,
are suspected of these diseases,

00:00:55.616 --> 00:00:58.243
an expert physician first orders

00:00:58.267 --> 00:01:00.893
very expensive medical
imaging technologies

00:01:00.917 --> 00:01:04.668
such as fluorescent imaging, CTs,
MRIs, to be performed.

00:01:05.046 --> 00:01:07.339
Once those images are collected,

00:01:07.363 --> 00:01:12.107
another expert physician then diagnoses
those images and talks to the patient.

00:01:12.522 --> 00:01:15.958
As you can see, this is a very
resource-intensive process,

00:01:15.982 --> 00:01:20.418
requiring both expert physicians,
expensive medical imaging technologies,

00:01:20.442 --> 00:01:23.521
and is not considered practical
for the developing world.

00:01:23.545 --> 00:01:27.037
And in fact, in many
industrialized nations, as well.

00:01:27.784 --> 00:01:31.069
So, can we solve this problem
using artificial intelligence?

00:01:31.887 --> 00:01:35.903
Today, if I were to use traditional
artificial intelligence architectures

00:01:35.927 --> 00:01:37.133
to solve this problem,

00:01:37.157 --> 00:01:38.736
I would require 10,000 --

00:01:38.760 --> 00:01:42.664
I repeat, on an order of 10,000
of these very expensive medical images

00:01:42.688 --> 00:01:44.085
first to be generated.

00:01:44.109 --> 00:01:46.974
After that, I would then go
to an expert physician,

00:01:46.998 --> 00:01:49.514
who would then analyze
those images for me.

00:01:49.538 --> 00:01:51.696
And using those two pieces of information,

00:01:51.720 --> 00:01:55.299
I can train a standard deep neural network
or a deep learning network,

00:01:55.323 --> 00:01:57.459
to provide patient's diagnosis.

00:01:57.483 --> 00:01:59.198
Similar to the first approach,

00:01:59.222 --> 00:02:01.388
traditional artificial
intelligence approaches

00:02:01.412 --> 00:02:02.856
suffer from the same problem.

00:02:02.880 --> 00:02:07.322
Large amounts of data, expert physicians,
and expert medical imaging technologies.

00:02:08.354 --> 00:02:12.601
So can we invent more scalable, effective,

00:02:12.625 --> 00:02:15.934
and more valuable
artificial intelligence architectures

00:02:15.958 --> 00:02:18.692
to solve these very important
problems facing us today?

00:02:19.034 --> 00:02:22.049
And this is exactly what my group
at MIT Media Lab does.

00:02:22.375 --> 00:02:26.200
We have invented a variety
of unorthodox AI architectures

00:02:26.224 --> 00:02:29.426
to solve some of the most important
challenges facing us today

00:02:29.450 --> 00:02:31.902
in medical imaging and clinical trials.

00:02:32.482 --> 00:02:35.537
In the example I shared
with you today, we had two goals.

00:02:35.561 --> 00:02:38.522
Our first goal was to reduce
the number of images

00:02:38.546 --> 00:02:41.815
required to train artificial
intelligence algorithms.

00:02:41.839 --> 00:02:43.926
Our second goal -- we were more ambitious,

00:02:43.950 --> 00:02:47.712
we wanted to reduce the use
of expensive medical imaging technologies

00:02:47.736 --> 00:02:48.934
to screen patients.

00:02:48.958 --> 00:02:50.292
So how did we do it?

00:02:50.966 --> 00:02:52.149
For our first goal,

00:02:52.173 --> 00:02:54.220
instead of starting
with tens and thousands

00:02:54.244 --> 00:02:57.253
of these very expensive medical images,
like traditional AI,

00:02:57.277 --> 00:02:59.355
we started with a single medical image.

00:02:59.379 --> 00:03:03.125
From this image, my team and I
figured out a very clever way

00:03:03.149 --> 00:03:05.895
to extract billions
of information packets.

00:03:05.919 --> 00:03:09.628
These information packets
included colors, pixels, geometry

00:03:09.652 --> 00:03:12.180
and rendering of the disease
on the medical image.

00:03:12.204 --> 00:03:16.537
In a sense, we converted one image
into billions of training data points,

00:03:16.561 --> 00:03:19.544
massively reducing the amount of data
needed for training.

00:03:20.148 --> 00:03:21.323
For our second goal,

00:03:21.347 --> 00:03:25.220
to reduce the use of expensive medical
imaging technologies to screen patients,

00:03:25.244 --> 00:03:28.093
we started with a standard,
white light photograph,

00:03:28.117 --> 00:03:32.482
acquired either from a DSLR camera
or a mobile phone, for the patient.

00:03:32.506 --> 00:03:34.974
Then remember those
billions of information packets?

00:03:34.998 --> 00:03:38.498
We overlaid those from
the medical image onto this image,

00:03:38.522 --> 00:03:41.474
creating something
what we call a composite image.

00:03:41.498 --> 00:03:44.776
Much to our surprise,
we only required 50 --

00:03:44.800 --> 00:03:46.133
I repeat, only 50 --

00:03:46.157 --> 00:03:50.022
of these composite images to train
our algorithms to high efficiencies.

00:03:50.704 --> 00:03:52.029
To summarize our approach,

00:03:52.053 --> 00:03:55.220
instead of using 10,000
very expensive medical images,

00:03:55.244 --> 00:03:58.252
we can now train the AI algorithms
in an unorthodox way,

00:03:58.276 --> 00:04:02.529
using only 50 of these high-resolution,
but standard photographs,

00:04:02.553 --> 00:04:05.070
acquired from DSLR cameras
and mobile phones

00:04:05.094 --> 00:04:06.617
and provide diagnosis.

00:04:06.641 --> 00:04:07.807
More importantly,

00:04:07.831 --> 00:04:11.029
our algorithms can accept,
in the future and even right now,

00:04:11.053 --> 00:04:13.840
some very simple, white light
photographs from the patient,

00:04:13.864 --> 00:04:16.593
instead of expensive medical
imaging technologies.

00:04:17.165 --> 00:04:20.228
I believe that we are poised
to enter an era

00:04:20.252 --> 00:04:22.167
where artificial intelligence

00:04:22.191 --> 00:04:24.720
is going to make an incredible
impact on our future.

00:04:24.744 --> 00:04:27.202
And I think that thinking
about traditional AI,

00:04:27.226 --> 00:04:29.998
which is data-rich but application-poor,

00:04:30.022 --> 00:04:31.569
we should also continue thinking

00:04:31.593 --> 00:04:34.601
about unorthodox artificial
intelligence architectures,

00:04:34.625 --> 00:04:36.569
which can accept small amounts of data

00:04:36.593 --> 00:04:39.539
and solve some of the most important
problems facing us today,

00:04:39.563 --> 00:04:40.801
especially in healthcare.

00:04:40.825 --> 00:04:42.000
Thank you very much.

00:04:42.024 --> 00:04:45.833
(Applause)