20.03.2021 Views

Deep-Learning-with-PyTorch

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Towards the finish line

405

14.1 Towards the finish line

We can get a hint of the work remaining by looking at figure 14.1. In step 3 (grouping)

we see that we still need to build the bridge between the segmentation model from

chapter 13 and the classifier from chapter 12 that will tell us whether what the segmentation

network found is, indeed, a nodule. On the right is step 5 (nodule analysis and

diagnosis), the last step to the overall goal: seeing whether a nodule is cancer. This is

another classification task; but to learn something in the process, we’ll take a fresh

angle at how to approach it by building on the nodule classifier we already have.

Step 1 (ch. 10):

Data Loading

STep 2 (ch. 13):

Segmentation

Step 4 (ch. 11+12):

ClaSsification

.MHD

.RAW

CT

Data

segmentation

model

Step 3 (ch. 14):

Grouping

candidate

Locations

[(I,R,C),

(I,R,C),

(I,R,C),

...

]

candidate

Sample

ClaSsification

model

[NEG,

p=0.1

POS,

p=0.9

NEG,

p=0.2

...

]

Step 5 (ch. 14):

Nodule analysis

and Diagnosis

MAL/BEN

p=0.9

Figure 14.1 Our end-to-end lung cancer detection project, with a focus on this chapter’s

topics: steps 3 and 5, grouping and nodule analysis

Of course, these brief descriptions and their simplified depiction in figure 14.1 leave

out a lot of detail. Let’s zoom in a little with figure 14.2 and see what we’ve got left to

accomplish.

As you can see, three important tasks remain. Each item in the following list corresponds

to a major line item from figure 14.2:

1 Generate nodule candidates. This is step 3 in the overall project. Three tasks go

into this step:

a Segmentation—The segmentation model from chapter 13 will predict if a

given pixel is of interest: if we suspect it is part of a nodule. This will be done

per 2D slice, and every 2D result will be stacked to form a 3D array of voxels

containing nodule candidate predictions.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!