@@ -129,7 +129,7 @@ def __call__(self,
129129 Returns:
130130 reshaped_data (np.ndarray): inquiry data of shape (Channels, Inquiries, Samples)
131131 labels (np.ndarray): integer label for each inquiry. With `trials_per_inquiry=K`,
132- a label of [0, K-1] indicates the position of `target_label`, or label of K indicates
132+ a label of [0, K-1] indicates the position of `target_label`, or label of [0 ... 0] indicates
133133 `target_label` was not present.
134134 reshaped_trigger_timing (List[List[int]]): For each inquiry, a list of the sample index where each trial
135135 begins, accounting for the prestim buffer that may have been added to the front of each inquiry.
@@ -229,6 +229,111 @@ def extract_trials(
229229 return np .stack (new_trials , 1 ) # C x T x S
230230
231231
232+ class GazeReshaper :
233+ def __call__ (self ,
234+ inq_start_times : List [float ],
235+ target_symbols : List [str ],
236+ gaze_data : np .ndarray ,
237+ sample_rate : int ,
238+ symbol_set : List [str ],
239+ channel_map : List [int ] = None ,
240+ ) -> dict :
241+ """Extract inquiry data and labels. Different from the EEG inquiry, the gaze inquiry window starts with
242+ the first flicker and ends with the last flicker in the inquiry. Each inquiry has a length of ~3 seconds.
243+ The labels are provided in the target_symbols list. It returns a Dict, where keys are the target symbols and
244+ the values are inquiries (appended in order of appearance) where the corresponding target symbol is prompted.
245+ Optional outputs:
246+ reshape_data is the list of data reshaped into (Inquiries, Channels, Samples), where inquirires are appended
247+ in chronological order. labels returns the list of target symbols in each inquiry.
248+
249+ Args:
250+ inq_start_times (List[float]): Timestamp of each event in seconds
251+ target_symbols (List[str]): Prompted symbol in each inquiry
252+ gaze_data (np.ndarray): shape (channels, samples) eye tracking data
253+ sample_rate (int): sample rate of data provided in eeg_data
254+ channel_map (List[int], optional): Describes which channels to include or discard.
255+ Defaults to None; all channels will be used.
256+
257+ Returns:
258+ data_by_targets (dict): Dictionary where keys are the symbol set and values are the appended inquiries
259+ for each symbol. dict[Key] = (np.ndarray) of shape (Channels, Samples)
260+
261+ reshaped_data (List[float]) [optional]: inquiry data of shape (Inquiries, Channels, Samples)
262+ labels (List[str]) [optional] : Target symbol in each inquiry.
263+ """
264+ if channel_map :
265+ # Remove the channels that we are not interested in
266+ channels_to_remove = [idx for idx , value in enumerate (channel_map ) if value == 0 ]
267+ gaze_data = np .delete (gaze_data , channels_to_remove , axis = 0 )
268+
269+ # Find the value closest to (& greater than) inq_start_times
270+ gaze_data_timing = gaze_data [- 1 , :].tolist ()
271+
272+ start_times = []
273+ for times in inq_start_times :
274+ temp = list (filter (lambda x : x > times , gaze_data_timing ))
275+ if len (temp ) > 0 :
276+ start_times .append (temp [0 ])
277+
278+ triggers = []
279+ for val in start_times :
280+ triggers .append (gaze_data_timing .index (val ))
281+
282+ # Label for every inquiry
283+ labels = target_symbols
284+
285+ # Create a dictionary with symbols as keys and data as values
286+ # 'A': [], 'B': [] ...
287+ data_by_targets = {}
288+ for symbol in symbol_set :
289+ data_by_targets [symbol ] = []
290+
291+ window_length = 3 # seconds, total length of flickering after prompt for each inquiry
292+
293+ reshaped_data = []
294+ # Merge the inquiries if they have the same target letter:
295+ for i , inquiry_index in enumerate (triggers ):
296+ start = inquiry_index
297+ stop = int (inquiry_index + (sample_rate * window_length )) # (60 samples * 3 seconds)
298+ # Check if the data exists for the inquiry:
299+ if stop > len (gaze_data [0 , :]):
300+ continue
301+
302+ reshaped_data .append (gaze_data [:, start :stop ])
303+ # (Optional) extracted data (Inquiries x Channels x Samples)
304+
305+ # Populate the dict by appending the inquiry to the correct key:
306+ data_by_targets [labels [i ]].append (gaze_data [:, start :stop ])
307+
308+ # After populating, flatten the arrays in the dictionary to (Channels x Samples):
309+ for symbol in symbol_set :
310+ if len (data_by_targets [symbol ]) > 0 :
311+ data_by_targets [symbol ] = np .transpose (np .array (data_by_targets [symbol ]), (1 , 0 , 2 ))
312+ data_by_targets [symbol ] = np .reshape (data_by_targets [symbol ], (len (data_by_targets [symbol ]), - 1 ))
313+
314+ # Note that this is a workaround to the issue of having different number of targetness in
315+ # each symbol. If a target symbol is prompted more than once, the data is appended to the dict as a list.
316+ # Which is why we need to convert it to a (np.ndarray) and flatten the dimensions.
317+ # This is not ideal, but it works for now.
318+
319+ # return np.stack(reshaped_data, 0), labels
320+ return data_by_targets
321+
322+ @staticmethod
323+ def centralize_all_data (data , symbol_pos ):
324+ """ Using the symbol locations in matrix, centralize all data (in Tobii units).
325+ This data will only be used in certain model types.
326+ Args:
327+ data (np.ndarray): Data in shape of num_channels x num_samples
328+ symbol_pos (np.ndarray(float)): Array of the current symbol posiiton in Tobii units
329+ Returns:
330+ data (np.ndarray): Centralized data in shape of num_channels x num_samples
331+ """
332+ for i in range (len (data )):
333+ data [i ] = data [i ] - symbol_pos
334+ return data
335+
336+
232337class TrialReshaper (Reshaper ):
233338 def __call__ (self ,
234339 trial_targetness_label : list ,
0 commit comments