Class Notes (943,161)
CA (551,708)
Western (52,351)
Lecture 5

Medical Biophysics 3503G Lecture 5: Medical biophysics lec 3

3 Pages
80 Views

Department
Medical Biophysics
Course Code
Medical Biophysics 3503G
Professor
James Lacefield

This preview shows page 1. Sign up to view the full 3 pages of the document.
Aside 1: MD terminology
Sensitivity = true positive fraction
Specificity = true negative fraction (TNF)
TNF = 1.0 – FPF
Overdiagnosis
In context of screening mammography, have exam that produces an overdiagnosis a false-
positive result or a true-positive result?
It’s true +ve in over diagnosed patients you said you found something by imaging test, so
something was really there, so was a true result
Overdiagnosis when you found tumour that would not have progressed to metasiss but was
treated anyway, so get side effects of treatment without benefits
TPF is in the dominator, all the cases that were cancer present by gold standard and in the
numerator is the number diagnosed properly
Fpf  total number of patients that were no cancer
ROC procedure
Now must reconsider, all possible plasces can put threshold and compure new tpf and fpf
changes diagnoiss for patients near threshold
You could put threshold at begiing of chart and say everyone has cancer this would mean all
red bars, are above threshold and got those people right, and therefore tpf is 1…all blue bars
are above threshold, saying people who don’t have cnacer have it, so FPF is also 1
Construct empirical ROC curve
plot (FPF, TPF) pairs obtained at each threshold
each point is a row of FPF and TPF from the table
if you’re doing this experimentally, tend to connect each of data pint with straight line, and we
flag unrealisticness by calling it empirical curve
we are determining receiver operating characteristic for imaging system
when we present the result graphically, the graph is called ROC curve
empirical prt is we connect our measured points with straight lines
Chance line
the red-dashed line is called the chance line (TPF = FPF) accuracy expected fi diagnosies
assigned randomly
this is ROC curve you would get if you assigned diagnosis randomly, instead of clinical
observation
find more resources at oneclass.com
find more resources at oneclass.com

Loved by over 2.2 million students

Over 90% improved by at least one letter grade.

Leah — University of Toronto

OneClass has been such a huge help in my studies at UofT especially since I am a transfer student. OneClass is the study buddy I never had before and definitely gives me the extra push to get from a B to an A!

Leah — University of Toronto
Saarim — University of Michigan

Balancing social life With academics can be difficult, that is why I'm so glad that OneClass is out there where I can find the top notes for all of my classes. Now I can be the all-star student I want to be.

Saarim — University of Michigan
Jenna — University of Wisconsin

As a college student living on a college budget, I love how easy it is to earn gift cards just by submitting my notes.

Jenna — University of Wisconsin
Anne — University of California

OneClass has allowed me to catch up with my most difficult course! #lifesaver

Anne — University of California
Description
Aside 1: MD terminology Sensitivity = true positive fraction Specificity = true negative fraction (TNF) TNF = 1.0 – FPF Overdiagnosis In context of screening mammography, have exam that produces an overdiagnosis a false- positive result or a true-positive result? It’s true +ve in over diagnosed patients you said you found something by imaging test, so something was really there, so was a true result Overdiagnosis when you found tumour that would not have progressed to metasiss but was treated anyway, so get side effects of treatment without benefits TPF is in the dominator, all the cases that were cancer present by gold standard and in the numerator is the number diagnosed properly Fpf total number of patients that were no cancer ROC procedure Now must reconsider, all possible plasces can put threshold and compure new tpf and fpf changes diagnoiss for patients near threshold You could put threshold at begiing of chart and say everyone has cancer this would mean all red bars, are above threshold and got those people right, and therefore tpf is 1…all blue bars are above threshold, saying people who don’t have cnacer have it, so FPF is also 1 Construct empirical ROC curve plot (FPF, TPF) pairs obtained at each threshold each point is a row of FPF and TPF from the table if you’re doing this experimentally, tend to connect each of data pint with straight line, and we flag unrealisticness by calling it empirical curve we are determining receiver operating characteristic for imaging system when we present the result graphically, the graph is called ROC curve empirical prt is we connect our measured points with straight lines Chance line the red-dashed line is called the chance line (TPF = FPF) accuracy expected fi diagnosies assigned randomly this is ROC curve you would get if you assigned diagnosis randomly, instead of clinical observationchance line is like flipping coin to make a diagnosis, flipping coin is gives you one point we flip a coint what’s the chance we get it right, it’s 1/2 , have 50% to get a heads, and will say that’s positive so TPF is 0.5 next patient comes into door doesn’t have cancer, flip coin, whats the chaince to get them wrong, we also have have 50% chance of calling that paitetnt positive, so FPF is 0.5 this is what you expect to get if you assigne diagnosis, not by any clinical observation, but by making disecsion using something that has nothing to do with health of patient, ex. what I rolled on the dice The other commone misperception is the chance line means you have 50% chance of being correct not true you must work out possibilities, where on chance line set decision threshold, and how many patients in population do or don’t have disease Perfect detector is the orange solid curve all roc curve
More Less
Unlock Document


Only page 1 are available for preview. Some parts have been intentionally blurred.

Unlock Document
You're Reading a Preview

Unlock to view full version

Unlock Document

Log In


OR

Don't have an account?

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


OR

By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.


Submit