File size: 11,496 Bytes
40d081b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
# Medical/Anatomical Data Visualization Test Cases for SciVisAgentBench
# Tests scalar volume visualization capabilities for medical imaging data

# Test 1: Basic Volume Rendering and Tissue Identification
- vars:
    question: |
      Clear the ParaView pipeline and load the data file "foot/data/foot_256x256x256_uint8.raw".
      1. Enable volume rendering to visualize the internal structures
      2. Adjust the opacity transfer function to reveal both bone and soft tissue (bone should be more opaque, soft tissue semi-transparent)
      3. Set an appropriate color map to differentiate tissue types (e.g., white/beige for bone, reddish for soft tissue)
      Finally, save the paraview state as "foot/results/{agent_mode}/foot.pvsm"
  assert:
    - type: llm-rubric
      value: |
        - Successfully load the foot dataset
        - Enable volume rendering
        - Adjust opacity to show both bone and soft tissue structures
        - Apply appropriate color mapping for tissue differentiation
        - Use screenshot to verify foot bones (metatarsals, phalanges) and soft tissue are visible
        - Verify the colors match the instruction (white/beige for bone, reddish for soft tissue)

# # Test 2: Multi-Isosurface Segmentation
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "mri_ventricles/data/mri_ventricles_256x256x124_uint8.raw".
#       1. Create at least 3 different isosurfaces at different threshold values to segment different tissue types
#       2. Color each isosurface differently to distinguish structures
#       3. Make appropriate surfaces semi-transparent if needed to show internal structures
#       Finally, save the paraview state as "mri_ventricles/results/{agent_mode}/mri_ventricles.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |
#         - Create multiple isosurfaces (at least 3) at different threshold values
#         - Apply different colors to each isosurface for clear distinction
#         - Use screenshot to verify brain ventricles are successfully segmented and visible
#         - Verify transparency is applied where needed to show nested structures
#         - Report identification of brain structures (ventricles, grey/white matter boundaries)

# # Test 3: Cross-Sectional Analysis
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "skull/data/skull_256x256x256_uint8.raw".
#       1. Create three orthogonal slices (axial, sagittal, and coronal planes)
#       2. Position the slices to show key anatomical features of the skull
#       3. Apply an appropriate color map to the slices
#       Finally, save the paraview state as "skull/results/{agent_mode}/skull.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Create exactly three orthogonal slices (axial, sagittal, coronal)
#         - Use screenshot to verify all three slice planes are visible simultaneously
#         - Verify slices show key features: cranial cavity, eye sockets, nasal cavity
#         - Report which anatomical features are visible in each specific plane

# # Test 4: Vascular Structure Visualization
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "aneurism/data/aneurism_256x256x256_uint8.raw".
#       1. Use appropriate visualization technique to isolate and display the vascular structure
#       2. Create an isosurface that clearly shows the aneurysm and blood vessels
#       3. Apply a red or red-gradient color map appropriate for vascular visualization
#       4. Compute and report the surface area of the vascular structure
#       Finally, save the paraview state as "aneurism/results/{agent_mode}/aneurism.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Successfully isolate vascular structure using isosurface
#         - Use screenshot to verify aneurysm bulge is clearly visible
#         - Verify red/red-gradient coloring is applied to vessels
#         - Successfully compute and report numerical surface area value

# # Test 5: Histogram Analysis and Tissue Classification
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "pancreas/data/pancreas_240x512x512_int16.raw".
#       1. Generate a histogram of the intensity values with 256 bins
#       2. Based on the histogram, identify distinct peaks corresponding to different tissue types
#       3. Create threshold filters to isolate pancreatic tissue based on the histogram analysis
#       4. Apply volume rendering with opacity settings based on your histogram findings
#       Finally, save the paraview state as "pancreas/results/{agent_mode}/pancreas.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Successfully generate histogram with 256 bins
#         - Report specific intensity values for identified peaks
#         - Create threshold filters using values derived from histogram peaks
#         - Apply volume rendering with opacity based on histogram analysis
#         - Use screenshot to verify pancreatic tissue is properly isolated

# # Test 6: Complex Organ System Visualization
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "mri_woman/data/mri_woman_256x256x109_uint16.raw".
#       1. Enable volume rendering to show internal anatomy
#       2. Adjust the transfer functions to visualize multiple organ systems simultaneously
#       3. Create a clip plane to show a sagittal cross-section while maintaining volume rendering
#       4. Use appropriate color and opacity settings to distinguish between organs
#       Finally, save the paraview state as "mri_woman/results/{agent_mode}/mri_woman.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Enable volume rendering successfully
#         - Adjust transfer functions to reveal multiple organ systems
#         - Create clip plane in sagittal orientation
#         - Use screenshots from anterior and lateral views to verify organ visibility
#         - List identified organs (e.g., heart, lungs, liver, spine)
#         - Verify clip plane and volume rendering work together

# # Test 7: Small Animal Specimen Analysis
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "frog/data/frog_256x256x44_uint8.raw".
#       1. Create visualization showing both skeletal and soft tissue structures
#       2. Use either multiple isosurfaces or volume rendering with careful transfer functions
#       3. Generate a plot-over-line measurement through the specimen (dorsal to ventral)
#       Finally, save the paraview state as "frog/results/{agent_mode}/frog.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Choose appropriate technique (isosurfaces or volume rendering)
#         - Use screenshot to verify both skeleton and soft tissue are visible
#         - Successfully create plot-over-line from dorsal to ventral
#         - Report numerical values from plot-over-line measurement
#         - Verify measurement traverses through the specimen correctly

# # Test 8: Dental Structure Analysis
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "tooth/data/tooth_103x94x161_uint8.raw".
#       1. Create isosurface to show the tooth enamel (outer layer)
#       2. Use volume rendering or additional isosurface to show internal structures (dentin, pulp cavity)
#       3. Apply appropriate colors (white for enamel, yellow for dentin)
#       Finally, save the paraview state as "tooth/results/{agent_mode}/tooth.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Create isosurface at appropriate threshold for enamel
#         - Use screenshot to verify white color on enamel and yellow/amber color on dentin
#         - Identify enamel, dentin, and pulp cavity structures

# # Test 9: Abdominal CT with Vascular Stent Visualization
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "stent/data/stent_512x512x174_uint16.raw".
#       This is an abdominal/pelvic CT scan containing a stent in the abdominal aorta (no contrast agent used).
#       1. Create a volume rendering to show the overall abdominal anatomy
#       2. Adjust the opacity transfer function to visualize bones (spine, pelvis) and soft tissue
#       3. Create an isosurface at a high threshold value to isolate and highlight the metallic stent
#       4. Apply appropriate colors - bone white/beige, soft tissue reddish, metallic grey for the stent
#       Finally, save the paraview state as "stent/results/{agent_mode}/stent.pvsm"

#   assert:
#     - type: llm-rubric
#       value: |

#         - Create volume rendering of full abdomen/pelvis
#         - Use screenshot to verify spine and pelvis are visible
#         - Create high-threshold isosurface to isolate metallic stent
#         - Verify stent appears with metallic/grey coloring
        
# # # Test 10: Comparative Analysis with State Saving
# # - vars:
# #     question: |
# #       Clear the ParaView pipeline and load the data file "vis_male/data/vis_male_128x256x256_uint8.raw".
# #       1. Create two different visualization approaches:
# #          a) Volume rendering optimized for soft tissue
# #          b) Isosurface extraction for skeletal system
# #       2. Save the ParaView state file to preserve your visualization setup
# #       3. Create a text report comparing the effectiveness of both techniques
# #   assert:
# #     - type: llm-rubric
# #       value: |
# # 
# #         - Create volume rendering with soft tissue emphasis
# #         - Create separate isosurface for skeletal system
# #         - Use screenshots to document both approaches
# #         - Successfully save .pvsm state file
# #         - Write and save comparative analysis as text file
# #         - Report specific advantages/disadvantages of each technique
# #         - Verify both visualizations show distinct anatomical features

# # Test 11: Marine Biology Specimen
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "lobster/data/lobster_301x324x56_uint8.raw".
#       1. Create visualization showing the exoskeleton and internal anatomy
#       2. Adjust the visualization to show the prominent structure
#       Finally, save the paraview state as "lobster/results/{agent_mode}/lobster.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - Choose appropriate visualization for exoskeleton (isosurface or volume)
#         - Use screenshot to verify whether specific features can be identified: claws, abdomen, tail fan

# # Test 12: Fish Anatomy Visualization  
# - vars:
#     question: |
#       Clear the ParaView pipeline and load the data file "carp/data/carp_256x256x512_uint16.raw".
#       1. Enable volume rendering to show internal fish anatomy
#       2. Adjust transfer functions to reveal skeletal structure and organs
#       3. Create a sagittal slice through the fish body
#       4. Use color mapping that differentiates tissue types
#       5. Reset camera with appropriate padding to frame the entire fish
#       Finally, save the paraview state as "carp/results/{agent_mode}/carp.pvsm"
#   assert:
#     - type: llm-rubric
#       value: |

#         - The volume rendering show both skeleton and organs
#         - Create sagittal slice along fish body length
#         - Verify sagittal slice shows internal structure along body axis