This
August
we
celebrated
successfully
becoming
registered
as
a
non-profit
cooperation
with
the
Texas
Secretary
of
State!
With
this
step
under
our
belt
we
can
officially
start
the
process
in
applying
for
our
501(c)(3)
status
with
the
IRS.
Research
and
Experiments
Experiment
Three:
Automated
Fetal
Cardiac
Segmentation
in
Ultrasound
Images We
are
excited
to
share
an
update
on
our
research
into
fetal
cardiac
segmentation,
a
crucial
step
in
advancing
prenatal
care
for
congenital
heart
defects.
Our
latest
experiment
focuses
on
developing
an
AI-powered
system
that
can
assist
clinicians
by
automatically
analyzing
fetal
heart
ultrasound
images
for
key
metrics
typically
conducted
manually
by
an
expert.
The
foundation
of
this
research
is
a
publicly
available
resource
known
as
the
FOCUS
dataset,
which
contains
fetal
ultrasound
images
that
have
been
meticulously
annotated
by
an
expert
sonographer.
These
annotations
function
as
a
"teaching
guide"
for
our
AI
models,
showing
them
the
precise
boundaries
of
key
anatomical
structures
within
the
fetal
heart
and
chest.
By
training
our
AI
to
identify
and
classify
these
details,
we
are
teaching
it
what
a
normal,
healthy
heart
should
look
like. This
trained
skill
is
critical
for
anomaly
detection.
Instead
of
a
clinician
manually
calculating
these
metrics,
our
AI
model
can
now
analyze
an
ultrasound
image
and
pinpoint
potential
deviations
from
the
norm.
This
allows
us
to
improve
the
accuracy
and
efficiency
of
current
screening
methods,
potentially
leading
to
the
earlier
detection
of
congenital
heart
disease
and
giving
families
more
time
to
prepare
for
life-saving
treatment.
Want
to
explore
more?
Dive
into
the
experiment
details
and
discover
fascinating
insights.
Showcasing
Our
Work
at
the
Kiro
Devpost
Hackathon We
are
excited
to
announce
that
Theodore's
Heart
Initiative
is
entering
the
Kiro
Devpost
Hackathon
with
a
new
application
designed
to
showcase
the
real-world
potential
of
our
research.
This
app
serves
as
a
proof
of
concept,
allowing
users
to
upload
ultrasound
images
for
analysis
and
anomaly
detection
using
the
same
models
we
have
been
training.
The
project
aims
to
demonstrate
how
our
work
could
be
utilized
by
clinicians
and
individuals
alike,
making
our
advanced
models
accessible
as
a
tool
for
preliminary
analysis.
By
participating
in
this
competition,
we
hope
to
achieve
two
key
goals.
First,
we
are
looking
to
generate
visibility
for
the
crucial
work
we
are
doing
to
advance
early
diagnosis
of
congenital
heart
disease.
Second,
any
portion
of
the
prize
pool
we
are
fortunate
enough
to
win
would
be
used
directly
to
fund
our
ongoing
research
and
development,
allowing
us
to
continue
creating
innovative
solutions
to
save
young
lives.
We
look
forward
to
sharing
our
progress
with
you!
Keep
in
mind
we
are
not
affiliated
with
the
Kiro
hackathon,
merely
participating
in
the
contest.
If
you
would
like
to
learn
more
about
the
details
and
the
Devpost
organization
as
a
whole
you
can
follow
the
link
below.
You
may
be
asked
to
create
an
account
to
browse
the
active
participants
of
the
Kiro
hackathon.
However,
this
isnβt
necessary
to
stay
informed
on
our
entry
as
weβll
include
relevant
updates
in
our
next
monthly
newsletter.
The
image
below
demonstrates
the
initial
process
for
selecting
the
model
for
analysis
and
showcases
the
pros
and
cons
of
each
current
version.
From
there
the
user
can
upload
an
image
that
will
be
processed
by
the
selected
model.
The
image
below
shows
the
results
of
an
uploaded
ultrasound
image
image.
This
displays
both
the
anomaly
detection
confidence
score
as
well
as
several
other
key
details
from
the
modelβs
analysis.
We
will
continue
to
make
updates
and
improvements
on
the
application
up
to
the
deadline
for
submission
of
the
contest
and
are
excited
to
be
able
to
be
able
to
showcase
our
research
in
a
tangible
way
through
this
competition!
We
are
so
incredibly
touched
by
all
of
the
generous
gifts
and
would
like
to
thank
you
all
for
the
continued
support.
Our
work
would
not
be
possible
without
the
gifts
we
have
received.
