TwinTree Insert


Chapter Fifteen
The MR Image
Processing and Visualization

15-01 Introduction

asically all digital images are pro­cessed in one kind or another. When reading MR images, one always should be pre­pared for the unexpected. Digi­tized imag­ing brings more of the unex­pected into our lives. Still, there are some easi­ly recognizable features in most images (Figure 15-01).

Figure 15-01:
Where and when were these pictures taken — and what do they show? Think twice … the answers are here.

Image processing was already on its way to becoming an esta­bli­shed field of re­­search in its own right when clinical MR imaging equipment be­came available in the ear­ly 1980s. The digital nature of MR imaging, cou­pled with a wide range of ap­pli­ca­tions, spurred an enormous activity in image-pro­cessing of MR imaging data in the last fifty years.

Computer assisted detection or diag­nosis (CAD) systems were de­ve­lo­ped to find and highlight suspicious regions or structures in images of the hu­man body by pat­tern recognition. They are employed in, e.g., the search for breast, bone, lung or meta­sta­tic cancer. These artificially intelligent (AI) systems can show astonishing per­for­man­ce. How­ever, they can also cause misclassification of images caused by hardly per­cep­tible per­turbations in the image data and thus be­come detrimental. The rea­son for such fail­ures are unknown [⇒ Szegedy 2014].

In this chapter we give a brief descrip­tion of image-processing techniques that have been applied to MR imaging. We also describe the somewhat related field of vi­su­­a­li­za­tion tech­ni­ques, with a special focus on 3D visualization methods. Some of the methods mentioned here relate directly to the next chapter on dy­na­mic imag­ing.

Involuntarily, image processing can add to the existing delusion and bias in image read­ing and lead to preconceived but wrong diagnoses (Figure 15-01 and Figure 16-01). Unrecognized optical and mental illusions caused by artifacts created by image-processing algorithms may amplify such errors (Figure 15-02). These prob­lems are beyond the scope of this introduction to MR imaging; they are treated in detail elsewhere [⇒ Frisby 1979].

Figure 15-02:
Top: The coffee house wall illusion — in rea­li­ty all lines are parallel.
Bottom: Fraser's spiral — in reality there are only cir­cles.

spaceholder redThe pictures on the top of the page … the answers to the questions: Where and when were these pictures taken — and what do they show?

These questions were just asked to confuse you. You might have thought the ans­wers are easy. Of course, they are.

(a) Where: A bird’s-eye view of Central Park in Manhattan. Wrong: This is a vodka com­mer­cial with a vodka bottle which looks like Central Park. This picture has been image-processed — and on this picture the East River is in the west, not in the east.
 Even if you believe that you know what you are seeing — think twice.

(b) When: Correct — before World War II (in 1928). Where: You are wrong — not in Chi­ca­go (even if there is a commercial for the Chicago Daily News), but in Berlin (at the corner of Unter den Linden boulevard and Friedrich Strasse in the city center).

Learning point: Always check the patient’s his­tory before you read your images and make a diagnosis — even when using AI support.