Download Clinical applications of diffusion

7/07/2012
CLINICAL APPLICATIONS OF DIFFUSIONWEIGHTED IMAGING IN THE ABDOMEN
Vincent Vandecaveye, Frederik De Keyzer
Department of Radiology
University Hospitals Leuven, Leuven, Belgium
ABDOMINAL DIFFUSION-WEIGHTED
IMAGING
Hepatic imaging
Extrahepatic imaging
•Non-cirrhotic liver
•Primary tumor detection
 lesion characterization
 detection of metastases
•Cirrhotic liver
 Tumor versus pseudo-tumor
 Prediction of lesion progression
 Transplantation?
 Pancreaticobiliary
 Endometrial carcinoma
 Uterine cervical carcinoma
 Intestinal tumors
 Prostate cancer
 Renal cancer
• Staging
 Lymph nodes
 Distant metastatic spread
Second primary tumors
Post-treatment imaging
•Abdominopelvic tumors
 post-surgical
 post-chemoradiation
Fibrosis versus tumoral recurrence
1
7/07/2012
differences in microstructure
Changes in H20 mobility
Facilitated diffusion
*Diffusion-weighted
image contrast
Restricted diffusion
-Mathematical quantification
Apparent diffusion coefficient
- No need for exogenic
contrast agent
- No irradiation  magnetic
-Worldwide -> not routine
- High expertise per centre
-Research
- Clinical routine
Native images: b-value (increasing the sensitivity for impeded diffusion)
Calculated ADC map
2
7/07/2012
DWI
optimum
slices
Distance factor
Scan position
Fase encodering
FoV read
FoV phase
Slice thickness
TR
TE
Averages
Concatenations
Fat Suppr
Fat sat mode
Base resolution
Phase resolution
Partial Fourier
Acceleration factor PE (SENSEGRAPPA)
Bandwidth
EPI factor
3T b-value
Minimal
31-36
31-36
0%
0%
• High
gradient strengths
isocenter
isocenter
• anterior-posterior
Field-strength? 1,5T vs 3Tanterior-posterior
* Minimal
380TE
mm
380 mm
* Maximum
Bandwidth
100%
81%
* Realistic
spatial resolution: tumor 5detection
5 mm
mm
* Volume
shim
6600
ms (anteroposterior)
5100
* Volume
67ofmsinterest + edge reserve84 ms
Edge of scan
3 volume increased artefacts
4
1
1
SPAIR
Fat Sat
strong
strong
192
128
80%
100%
7/8
6/8
2
1736Hz/Px
154
0
1502
104
Correct shim position: volume shim > auto shim
0,50,100,300,600,1000
0,50,100,300,600,1000
6 good reasons to use a high b-value (b600  b1000)
3
7/07/2012
CHOICE OF B-VALUE
Similar image quality high vs low b-value
Higher accuracy high vs low b-value
BASIC IMAGE INTERPRETATION: PRIMARY STAGING
•
Current criteria in our center  b0-b100-b600-b1000 (free breathing DWI):
• Lymph nodes: ADC  malignant < 0,001mm2/sec < benign
• Sens: 76-83%; spec 94%
• Colorectal, gastro-intestinal and ovarian
• Non-cirrhotic liver : malignant (b1000+) <0,0011<benign (b1000-/+)
T2 anatomical correlate
• Cirrhotic liver : b600 SI + anatomical imaging – contrast-enhanced
•
Skeletal/non-hepatic soft tissue metastasis : b1000+ / anatomical correlate
• Primary/tumor : b1000+/ anatomical correlate
• CAVE: pancreatic applications: anatomical correlate and ADC
Major cause for false positives: abscess, hematoma, granuloma – low ADC
4
7/07/2012
BASIC IMAGE INTERPRETATION: TUMOR RECURRENCE
End treatment TN2
Primary location: b1000 + (ADC) + anatomical correlate
Nodal disease: b1000 + ADC (TH 0,0014)
T2 SHINE THROUGH
False positive b1000
5
7/07/2012
LOW T2
INSUFFICIENT SIGNAL @ B0
False positive ADC
NO CONTRAST LESION
TO BACKGROUND
B1000 indeterminate
detection
Characterization
6
7/07/2012
INSUFFICIENT SOLID TUMOR COMPONENT
False negative ADC
B1000 heterogeneity
LIVER IMAGING
** Standardize interpretation criteria
- Qualitative and quantitative
- Native b-images > ADC
Diffusion-weighted anatomy:
- Normal condition
- Pathologic condition
** Artefacting
** Pitfalls
** Clinical indication
7
7/07/2012
LIVER IMAGING
- Both on clinical indication and methodology a distinct difference should be made between
the non-cirrhotic and cirrhotic liver
- Capture all morphologic features on native DWI images for characterization
- ADC is highly dependent of signal variations on b0-b1000
- Main indication in “non-cirrhotic” liver is staging (liver metastases)
- Main indication in “cirrhotic liver” is detection of HCC
- Consider benign lesions as “avoid a false positive”
- Should we really make a difference between lesion detection and characterization
- If you detect a liver metastasis you are very likely to call it a metastasis
NON-CIRRHOTIC LIVER: ANATOMY
b0
b0
b0
b100
b100
b100
b100
*
b1000
b1000
b1000
b1000
b600
b600
b600
b600
Left lobe:
Cardiac pulse
Base of lung
Signal loss at high b
* Segment 5 and 6:
adjacent hepatic flexure
of colon
-Signal loss
- distortion artefact
-Due to air tissue interface
*
+
* Bile  viscous fluid:
Prolonged SI on native DWI
8
7/07/2012
NON-CIRRHOTIC : DETECTION OF METASTASES
56 patients with primary tumor
Sens = 96,6%
Spec = 93%
B1000-SI
Combination of
b1000 signal intensity(SI)
ADC
Metastases: Native DWI b50-b1000
b1000 - => NPV > 90%
STOP
Cave necrotic metastases
ADCb0-1000
Native images signal heterogeneity !!!
b1000 + => ADC
=> T2(Te63/Te363)
Metastases – hemangioma – FNH/adenoma
SI
T2-TE 63
T2-TE 263
Hemangioma
ADC
b
b0
b50
b600
b1000
FNH
SI
ADC
b
b0
b50
b600
b1000
b50
b600
b1000 Adenoma
SI
ADC
b
b0
Metastases
SI
ADC
b0
b
b50
b600
b1000
9
7/07/2012
Benign lesions: cyst/hemangioma
Absent signal on b1000 or T2 – shine through - always ADC>0.00120
Cyst
 hemangioma
 Cave cavernous hemangioma: may be heterogeneous, may show areas of low ADC
 THUS: diagnosis should preferentially be made by early-late T2
 Value for DWI: avoid false positive – exclude M+ in subcentimetric lesions seen on CT
Benign lesions:
cyst/hemangioma
b1000
M+
10
7/07/2012
Solid benign lesions
FNH and adenoma have a highly variable appearance on DWI/ADC
Exploit every possible morpholgic feature
ADC-threshold: dependent on base signal on native images
Overlap with metastases of adenocarcinoma, but less signal on native DWI
Variable ADC and SI of lesions due to heteroheneity in histologic background:
Steatotic adenoma - FNH
Teleangiectatic / inflammatory adenoma….
SOLID BENIGN LESIONS: FNH
FNH
b0
b50
b600
b1000
FNH
b0
ADC b50
b600
b1000
If no signal on native b-values  consider benign – don t calculate ADC
ADC range: 0.0011 – 0.0015
Also check for morphologic clues
M+
b0
b50
b600
b1000
11
7/07/2012
SOLID BENIGN LESIONS:
THE VARIABLE APPEARANCE OF ADENOMA
b1000
Steatotic adenoma  if b1000 is negative  don t proceed with ADC, consider negative
b1000
b1000
ADC
ADC
Malignant lesions: metastases
B1000 Signal intensity and signal heterogeneity
Threshold UZ: ~ 0.00011
Correlate with native images
12
7/07/2012
MALIGNANT LESIONS: METASTASES
M+ breast cancer
M+ colorectal  major additional value = detection of millimetric metastases = better staging
CAVE necrotic and mucinous metastases: lesion heterogeneity – false negative ADC
IMAGING OF CIRRHOSIS
Screening: early detection of HCC  curability
Negative: follow-up
Positive
Staging
Transplantation
Anatomy
Cirrhosis
Complications
Systemisch
chemotherapie
Surgery - RFA
TACE
SIRS
Chem-lip
supportief
13
7/07/2012
Cirrhosis: diffusion-weighted anatomy
b0
b0
b0
b100
b100
b100
b600
b600
b1000
b1000
b1000
Iron overload
of liver
Fibrotic septa
Liver background in cirrhosis is most important pitfall
PATHOPHYSIOLOGY: CELLULAR CHANGES
Window of opportunity: early cellular
changes
Exclusion of microcirculatory pollution
B-value > 300
Cellular
Tumorperfusion
Hypovascular HCC
Hypervascular pseudotumor
14
7/07/2012
Cirrhosis image generation
Microperfusion
True diffusion
Microperfusion
True diffusion
True diffusion
True diffusion
ADC
ADC
b0
b300
b600
b1000
b300
b0
b1000
b600
Regenerative nodule
High grade dysplastic nodule/early HCC
Cirrhotic liver
Cirrhotic liver
Cirrhosis: signal generation
> 2cm
< 2cm
b600-
T2-CE
SIratio
MRI
True positive
28
27
31
23
2
2
7
16
9
9
34
25
0
1
3
11
Sensitivity
100%
96.4%
91.2%
67.6%
Specificity
81.8%
81.8%
82.9%
61.0%
3. Improved conspicuity of small lesions
Accuracy
94.9%
92.3%
86.7%
64.0%


PPV
93.3%
93.1%
81.6%
59.0%
NPV
100%
90.0%
91.9%
69.4%
1.
No ADC quantification :
False positive

Noise versus diffusion restriction
True negative
b600-SIratio
T2-CE
MRI
False
2. b600-SI optimal threshold


Cellular changes => malignant transformation
Exclude perfusion effects => pseudotumor
Early detection of HCC
Pretransplant staging?
4. Prediction of lesion progression


negative
V. Vandecaveye et al, Eur radiol 2009
Timing of follow-up/intervention
Eligibility for transplantation
15
7/07/2012
T2
T1
Art
Ven
b600
RN
LDN
HDN
Early HCC
HCC
Heterogeneity as a diagnostic marker?
- Anatomical lesion size does not correspond with DWI lesion size
- The hepatocellular paradox  usually large HCC is not visible at DWI
according to conventional standards
 small lesions easily identified
- Tool for characterization: yes – only in combination with contrast-enhanced imaging
- Staging tool due to the improved detection of subcentimeter lesions
- Large lesions: often well differentiated HCC – DWI only detects regions with highest
tumoral cell density
 DWI lesion-in-lesion
16
7/07/2012
Heterogeneity as a diagnostic marker
Well differentiated HCC
Detection and characterization
HCC
HCC
17
7/07/2012
Staging
Pitfalls
Low grade dysplastic nodule
b50
B600 Ratio = 1,3
HCC – fibrotic variant
Fatty degeneration
 Always correlate DW-MRI to conventional MRI  characterization
18
7/07/2012
Pitfalls
Nodular regeneration after portal vein thrombosis
Confluent fibrosis
Cholangiocarcinoma
19
7/07/2012
Cholangiocarcinoma
- The major role of DWI is for lesion detection – characterization of malignancy and staging
Currently unclear if the detection of cholangiocarcinoma can be improved on a background of PSC
- No specific DWI features that allow the pure differentiation from M+ or HCC  + contrast-enhanced MRI
b1000
Coronal reformatted b1000
Improved detection of small lesions - Cui XY, World J Gastroenterol 2010
Cholangiocarcinoma
Routine MRI-cholangiography (1.5 T) in patient with jaundice and right hypochondric pain
Cholangiocarcinoma
Free margin to left lobe (Klatskin IIIa)
Right extended hemihepatectomie planned
20
7/07/2012
3T
Cholangiocarcinoma
b1000 coronal reformat
?
Hilar and intrahepatic tumor infiltration
Peritoneal metastasen
Inoperable
1.5 mm
mm slice
slice thickness
thickness
33 mm
1.5
1.5 mm
mm slice
slice thickness
thickness
Extrahepatic imaging
IMAGING PANCREATIC CANCER: A
CONTROVERSY?
ABDOMINOPELVIC IMAGING / RECURRENCE
IMAGING
21
7/07/2012
PANCREATIC DWI: ANATOMY
- Highly variable depending on the presence of chronic pancreatitis, lipomatosis, etc….
b0
b100
b600
b1000
IMAGING PANCREATIC CANCER: A
CONTROVERSY?
•
•
adenocarcinoma
High b-value DWI/ADC shows high Pancreatic
accuracy for
pancreatic
adenocarcinoma
•
Ichikawa T, Abdominal Imaging 2007
•
Kartalis N, Eur Radiol 2009
More restriction
Pancreatic adenocarcinoma shows hyperintense signal on high b-value
due
to restriction
diffusion restriction
Less
• CAVE variable contrast with pancreatic carcinoma
Exocrine pancreas
• Anatomical imaging
22
7/07/2012
PANCREATIC DWI: INDICATIONS AND
INTERPRETATION
** Currently no clear agreement
UZ Leuven: indications and interpretation
** Detection of neuro-endocrine tumors
b1000 + anatomy
** Detection of carcinoma in chronic pancreatitis
b1000 + ADC + anatomy
** Facilitate detection of small adenocarcinoma in dilated Wirsung duct
small sized lesions: b1000+ anatomy
** Malignant transformation of IPMN
solid component in cyst?: b1000+anatomy
NEURO-ENDOCRINE TUMOR
Diagnosis of pancreatic neuro-endocrine tumor
23
7/07/2012
NEURO-ENDOCRINE TUMOR
Patient with repetitive hypoglycemic attacks
T1
T2
?
?
Arterial phase
b50
b1000
Chromogranine staining
Improved lesions conspicuity -> improved lesion detection
CARCINOMA IN CHRONIC
PANCREATITIS
24
7/07/2012
DUCTAL CHANGES
Facilitation of diffusion in inflammatory fibrotic tissue
MALIGNANT TRANSFORMATION OF
IPMN
25
7/07/2012
RESTAGING POST-TREATMENT
Tumoral recurrence:
Solid hypercellular
•b1000 +
B1000 -
•ADC often decreased
B1000 +
Surrounding tissue:
Inflammation-fibrosis
•b1000 –
•ADC often increased
Koh D, Collins D; AJR 2007
•Variable in late phase
TUMOR HETEROGENEITY
Viable rim
Necrotic center
26
7/07/2012
POST-TREATMENT IMAGING
Patient with elevation of CEA
after total mesorectal excision and
neo-adjuvant (chemo)radiotherapy
TSE-MRI
Fibrosis/inflammation versus tumour recurrence?
ADC=0,00105
Pelvic recurrence confirmed by surgical biopsy
RESTAGING POST-TREATMENT
Prior neo-adjuvant chemoradiation and surgery for rectal cancer
CT shows intrapelvic mass – normal tumormarker CEA
b1000
DWI as an additional imaging technique
Screen for restrictive lesions
Pelvic excenteration
tumor recurrence
Signal suppresion in inflammation
Application chemoradiation>surgery
27
7/07/2012
Patient with prior low anterior resection after neo-adjuvant chemoradiation for rectal cancer.
Clinically increasing tumor marker, Increased FDG-uptake at the site of resection.
b1000
ADC=0,00142
b1000
ADC= 0,00098
b1000
ADC= 0,00098
Pitfall: abscess !!!!! Major cause of false positive
Correlate to conventional imaging
Area of liquefaction -> restricted diffusion -> likely abcess
-> facilitated diffusion -> likely (tumoral) necrosis
28
7/07/2012
TAKE HOME MESSAGES
Investing into knowledge and
Investing
traininginto communication
29