Lung cancer detection system..
in [Matlab]
|
Prev: probability density function vs likelihood
Next: regarding Sensorless Control of Permanent Magnet Synchronous Motor Based on Extended Kalman Filterl
From: dpb on 22 May 2010 10:27 Gaya wrote: .... > sorry..I thought someone will reply me with their experince with a > similar project... Have you done literature searches, etc., ... I'd think there would be a "veritable plethora" of what has been used for such analyses previously widely available including how successful/unsuccessful they've been...(spoken as an engineer, not bio-field related at all). --
From: Gaya on 22 May 2010 10:45 dpb <none(a)non.net> wrote in message <ht8po5$bmn$2(a)news.eternal-september.org>... > Gaya wrote: > ... > > > sorry..I thought someone will reply me with their experince with a > > similar project... > > Have you done literature searches, etc., ... I'd think there would be a > "veritable plethora" of what has been used for such analyses previously > widely available including how successful/unsuccessful they've > been...(spoken as an engineer, not bio-field related at all). > > -- thank you..:-)
From: Gaya on 22 May 2010 10:47 "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message <ht8pnh$sdq$1(a)fred.mathworks.com>... > "Gaya " <gayaudeshani(a)yahoo.com> wrote in message <ht8p10$d3d$1(a)fred.mathworks.com>... > > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message <ht8i1s$l50$1(a)fred.mathworks.com>... > > > "Gaya " <gayaudeshani(a)yahoo.com> wrote in message <ht8dl7$bpn$1(a)fred.mathworks.com>... > > > > Dear all, > > > > > > > > I'm currently developing a cancer detection system. > > > > I use digital chest xray images as the input to the system. I extract 6 first order statistic texture features and 4 second order statistic features. > > > > > > > > Is that enought to consider these features as the input to train the neural network..? > > > > > > As Mark points out, this is not a matlab question. Nor can > > > anybody answer (on any newsgroup or forum) if you will > > > ever have ENOUGH data. The answer is that "enough" can > > > only be determined by you, knowing your goals. The more > > > data that you provide, the better in general will be your > > > results. > > > > > > John thank you > > > > Thank you > > but sometimes people can answer by their experience.. > > And sorry I am new to this newsgroup. I use Mathlab to develop this thatz why I asked that question... > > Again, our experience is not even relevant. We cannot > say if you have enough data. You will never know that > you have enough data. > > The more data you have, the better your estimates. > > John
From: Walter Roberson on 22 May 2010 11:06
Gaya wrote: > "John D'Errico" <woodchips(a)rochester.rr.com> wrote in message > <ht8i1s$l50$1(a)fred.mathworks.com>... >> "Gaya " <gayaudeshani(a)yahoo.com> wrote in message >> <ht8dl7$bpn$1(a)fred.mathworks.com>... >> > Dear all, >> > > I'm currently developing a cancer detection system. >> > I use digital chest xray images as the input to the system. I >> extract 6 first order statistic texture features and 4 second order >> statistic features. >> > > Is that enought to consider these features as the input to train >> the neural network..? >> >> As Mark points out, this is not a matlab question. Nor can >> anybody answer (on any newsgroup or forum) if you will >> ever have ENOUGH data. The answer is that "enough" can >> only be determined by you, knowing your goals. The more >> data that you provide, the better in general will be your >> results. > Thank you but sometimes people can answer by their experience.. Our experience: neural nets are only as good as the features you train them on, and figuring out which features to train on (without "over-training") is a difficult task. By the time you have calculated what to train on, you probably have a program that does not need or benefit from neural nets. > And sorry I am new to this newsgroup. I use Mathlab to develop this > thatz why I asked that question... If you drink coffee, then you are using coffee to develop the program, so why not ask in the alt.coffee newsgroup? Most people that develop such programs are coffee drinkers and not so many that seriously develop such programs use Matlab, so "logically" alt.coffee would be a better place to ask. |